The Reproductive Biology, Natural Enemies and Biological Control of odorata Lem.

by Carol Ann Rolando

Submitted in fulfilment of the requirements for the degree of Master of Science

in the

School of Botany and Zoology, University of Natal, Pietermaritzburg,

March 2000 ABSTRACT

Delairea odorata Lem., an asteraceous perennial indigenous to southern Africa, has become naturalised and invasive in many subtropical regions including California, South Australia and Hawaii. Biological control offers a potential long term solution to the management of this species in exotic locations. This study analysed aspects ofthe biology ofD. odorata in its native environment to determine its suitability to classical biological control. To this end an examination of the reproductive biology and natural enemies of D. odorata was made. A study of the pyrrolizidine alkaloid profile was also conducted.

Reproductive biology: Delairea odorata reproduces both sexually by seeds and asexually by stolons. The flowering season occurs over the autumn months from April to June. Results ofthe trials indicate thatD. odorata is a cross compatible species and an obligate outbreeder. There is no specialised pollination system and the predorninant pollinators belong to the families Apidae, Syrphidae and Calliphoridae. Following pollination, numerous small achenes are produced. Laboratory trials indicate that these achenes germinate readily between 10 and 25 QC and, although germination occurs in both the light and dark, light clearly stimulates seed germination. Greenhouse trials conducted to determine the effect of light on growth and reproduction indicate that D. odorata is a shade tolerant species which shows plasticity in terms ofgrowth form and deployment ofbiomass in response to changes in light intensity. Growth rate and allocation ofbiomass to vegetative and sexual reproduction are highest at an intermediate light level. However, greatest allocation ofbiomass is to stem growth regardless oflight level.

Natural enemies: Surveys for potential biological control agents against Delairea odorata were conducted in KwaZulu-Natal and several phytophagous species were associated with the . However, only one potentially suitable control agent was identified, a stem galling tephritid , regalis Munro. Preliminary studies indicate this species to be fairly host-specific, a valuable asset if it is to be considered as a control agent. Furthermore, as D. odorata proliferates extensively by means ofstem regeneration and elongation, galling ofthese growing points by P. regalis may limit stolon spread in exotic locations. Two species ofparasitic wasp (Braconidae) were found to parasitise P. regalis pupae. IfP. regalis is to be used as a control agent the likelihood ofparasitisation in the new environment must be determined. 11

Pyrrolizidine alkaloids: Host-specificity in is often dependent on host-plant chemistry (e.g. alkaloids or essential oils). Thus prior to any biological control programme it is important to determine ifthere are ecotypes ofthe host plant present. An investigation to determine the specificity ofthe pyrrolizidine alkaloid profile ofD. odorata, occurring across KwaZulu-Natal, was made. The results indicate the presence of nine retronecine based pyrrolizidine alkaloids which occur in similar proportions in locally distributed . However, these alkaloid profiles differ considerably from those published for D. odorata occurring in California. This is an interesting and important result which indicates that chemotypes ofD. odorata may exist, a factor which must be considered in the initiation ofany biocontrol. Ifchemotypes ofD. odorata are present this may affect the behaviour ofnatural control agents on the exotic plant populations. III

Preface

The experimental work in this thesis was carried out under the supervision ofDr T. Edwards, at the School ofZoology and Botany, University ofNatal, Pietermaritzburg from February 1998 to August 1999.

The results ofthis study have not been submitted in any form to another university and, except where acknowledged, are the results ofmy own work.

~~nn Rolando IV

Contents

page

Abstract , i Preface iii Acknowledgements vii List of Figures viii List of Tables xiv

Chapter 1. General introduction 1 1.1 Project description 1 1.2 Weeds: their definition, attributes and control 2 1.3 Biological control 5 1.4 General background to the present study ...... 11 1.5 Overall aims of the present study 19

Chapter 2. Weeds of the and their biological control: a review " ...... 21 2.1 Introduction 21 2.2 The database and methods used for this review 23 2.3 Biocontrol agents used in the control of the Asteraceae 24 2.4 Summary and conclusions 33

Chapter 3. Floral biology, breeding system and pollination...... 35 3.1 Introduction 35 3.2 Materials and methods 38 3.3 Results 41 3.4 Discussion 48 v

Chapter 4. Seed biology ...... 52 4.1 Introduction 52 4.2 Materials and methods 55 4.3 Results 57 4.4 Discussion 65 4.5 Concluding comments 68

Chapter 5. Growth and reproduction in relation to light availability .. 70 5.1 Introduction 70. 5.2 Materials and methods 73 5.3 Results 74 5.4 Discussion 82 5.5 Concluding comments 85

Chapter 6. Natural enemies of De/airea odorata and a potential biological control agent...... 86

Section 6.1. Natural enemies 86 6.1.1 Introduction 86 6.1 .2 Materials and methods 88 6.1.3 Results and discussion 88

Section 6.2. Studies on a potential biocontrol agent - Parafreutreta regalis Munro (Diptera: ) 92 6.2.1 Introduction 92 6.2.2 Materials and methods 92 6.2.3 Results and discussion 93 6.2.4 Concluding comments 109 VI

Chapter 7. Pyrrolizidine alkaloids 112 7.1 Introduction 112 7.2 Materials and methods 114 7.2.1 Development of an analytical technique 114 7.2.2 Methods used for the extraction, isolation and identification of alkaloids . 117 7.3 Results and discussion 118 7.4 Concluding comments 126

Chapter 8. Summary and conclusions 129

Chapter 9. References 133

Appendix A ...... 162 vu

Acknowledgements

I would like to thank my supervisor, Or T. Edwards, for the advice and encouragement which he gave me during the course of this investigation.

Special thanks must also go to: Prof. J. van Staden for both his academic and financial support for which I am extremely grateful.

Or R. M. Miller for his advice and help with identification of invertebrates.

Or M. Smith for his expertise, assistance and patience concerning the pyrrolizidine

alkaloid analysis.

Credit is also due to the following persons: Or E. Grobbelaar (PPRI) and Or R. Muller (Transvaal Museum) for their help with identification of .

OrM. Kruger (Transvaal Museum) for his assistance with the identification of and .

The staff of the Electron Microscope Unit for their help with scanning electron microscopy.

I would also like to thank my parents for their continued support and encouragement.

Finally, I would like to thank Hallam who provided the confidence, inspiration, motivation and support when I needed it - which was most of the time! Vlll

List of Figun!s

Figure 1.1 Distribution of Oelairea odorata in (Distribution map is based on specimens lodged at herbaria: PRE and NU) 14

Figure 1.2 Distribution of Oelairea odorata on the main Hawaiian Islands as of 1986 (HEAR, 1986) 15

Figure 3.1 Intact anther collar, of a 0. odorata floret, consisting of five - syngenecious anthers.. 42

Figure 3.2 Status of style when the O. odorata floret is still closed. Floret and anther collar have been prised open to show style apex positioned at the base of the anther collar 42

Figure 3.3 The style pushes through the narrow anther collar of the O. odorata florets and removes pollen out of the cylinder by means of hairs on the outer surface. . 43

Figure 3.4 After the style has extended beyond the anther collar, the style arms spread apart and expose the stigmatic surface. . 43

Figure 3.5 a-c Phenology of O. odorata at (a) Ferncliffe (b) Ingele and (c) Dargle ...... 44

Figure 3.6 a&b Results of experimental on 0. odorata capitula expressed as (a) the mean number offruits percapitulum (error bars represent 95 % confidence intervals) and (b) the percentage capitula which set fruit (numbers refer to the total number of capitula used in the test). . 46 IX

Figure 3.7 Mean number offruits per capitulum for pollen limitation experiments carried out on D. odorata capitula. Error bar represents 95 % confidence interval. HP: hand-pollinated capitula; EC: open pollinated capitula on experimental plants; CC: open pollinated capitula on control plants , 47

Figure 4.1 Increase in mass (water uptake) offruits of D. odorata imbibed at 25 "C for 12 days. Photoperiod was set at 16 h lighU8h dark. Error bars represent standard error 58

Figure 4.2 Cumulative % germination of D. odorata achenes maintained for 15 days at one of five temperatures. Photoperiod was set at 16 h IighU 8 h dark. Error bars represent standard error...... 59

Figure 4.3 Cumulative % germination of D. odorata achenes maintained for 20 days in the dark at one of five temperatures. Error bars represent standard error 59

Figure 4.4 Cumulative % germination of D. odorata achenes following treatment for one month in the dark at 0 "C. Achenes were germinated at 15°C in both the light and the dark. Error bars represent standard error 61

Figure 4.5 Final % germination, in the light and dark, of dry stored (1 yr) D. odorata achenes, following 20 days treatment at 10, 15 and 25°C. Error bars represent standard error. . 62

Figure 4.6 Emergence and subsequent survival (%) of D. odorata achenes following eight months treatment at 71,45 and 17 % sunlight. Error bars represent standard error ...... 63 x

Figure 4.7 Monthly increase in stem length (mm) of D. odorata seedlings exposed to 71,45 and 17 % sunlight. Error bars represent standard error 64

Figure 4.8 Monthly increase in numbe r of D. odorata seedlings exposed to 17, 45 and 71 % sunlight. Error bars represent standard error. 64

Figure 4.9 Length of the longest (mm) of D. odorata seedlings following eight months exposure to one of three light levels. Error bars represent standard error : 66

Figure 4.10 Dry weight allocation (%) to roots stems and leaves of D. odorata seedlings following eight months exposure to one of three light levels. Error bars represent standard error.. . 66

Figure 5.1 Mean leaf length vs. width (mm) of 10 mature leaves (n=60) of D. odorata plants exposed to one of three light levels...... 76

Figure 5.2 a&b Petiole length (a) and leaf number (b) of D. odorata transplants after 12 weeks exposure to one of three light levels. Horizontal bar represents the median. . 77

Figure 5.3 Mean length (cm) of the first 10 sequentially produced internodes along the primary stolons of D. odorata cuttings, following 12 weeks exposure to one of three light levels (n=60). Error bars represent standard error 79

Figure 5.4 a-d Dry weights for whole D. odorata plants and component plant parts after 12 weeks exposure to different light levels (n=28). Horizontal bar represents the median 80 Xl

Figure 5.5 Biomass allocation (%) to leaves, roots and stems by D. odorata cuttings grown at 71,45 or 17 % PAR (n=28). Error bars represent standard error 81

Figure 5.6 Stem biomass (%) allocated to secondary stolons by D. odorata cuttings grown at 71,45 or 17 % PAR (n=28). Error bars represent standard error. . 81

Figure 5.7 Total dry weight (g) of produced by D. odorata transplants following eight months treatment at one of three light levels 83

Figure 5.8 Dry weig ht (g) of inflorescences produced as a function of whole plant dry weight (g) 83

Figure 6.1 Parafreutreta rega/is (7.5X). . 94

Figure 6.2 Parafreutreta conferta (7X) 94

Figure 6.3 Parafreutreta felina (6.9X). . 94

Figure 6.4 a-d Data recorded from stem galls collected on D. odorata (n= 20), S. tamoides (n=6) and S. quinque/obus (n=12). All values represent the mean, error bars represent standard error 95

Figure 6.5 Fungal infected galls (12X). . 96

Figure 6.6 Sub-translucent P. rega/is puparia in the pith of a gall (14X) .. 98

Figure 6.7 Glossy black puparia typical of P. felina (12.5X) 98 XII

Figure 6.8 a&b Parasitic Hymenoptera (Braconidae) collected from D. odorata galls (7.5X) 99

Figure 6.9 a Weekly increase in length (mm) of developing D. odorata stem galls. · 102

Figure 6.9 b Weekly increase in width (mm) of developing D. odorata stem galls. · 102

Figure 6.10 The mature larvae scrapes away a spot on the wall leaving a window through which the young adult can emerge. Generally only one "peep window" is formed in the P. regalis galls (1 OX). . . 104

Figure 6.11 Transverse section (T/S) through a young D. odorata stem (150X) · 105

Figure 6.12 T/S of a mature D. odorata stem showing "anomalous secondary thickening" (175X) 105

Figure 6.13 T/S of a gall showing a tephritid larva feeding on nutritive tissue. Note disruption of collenchyma tissue and vascular bundles (185X) · 107

Figure 6.14 T/S of galled stem tissue showing the highly meristematic pith region and lateral stretching of cells (167X) ...... 107

Figure 6.15 T/S of stem tissue 3 cm below the gall (170X) 108

Figure 6.16 T/S of stem adjacent to the gall (when galling on axillary meristems)

showing the disruptive effect of the gall (165X) 108 Xlll

Figure 7.1 Separation of an alkaloid extract from De/airea odorata by GC ...... 119

Fiqure 7.2 GC-MS of alkaloid extract. 119

Figure 7.3 a-f Pyrrolizidine alkaloids (ug/g) found in flowers, leaves and stems of D. odorata plants collected at two sampling sites at three localities; Ingele (a&b), Dargle (c&d) and Ferncliffe (e&f) 124

Figure 7.4 Relative distribution of pyrrolizidine alkaloids within different D. odorata plant parts. Plants collected at two sampling sites at three localities: Ingele, Ferncliffe and Dargle 125

Figure 7.5 a-d Weight (mg/10 9 dry plant material) of pyrrolizidine alkaloid extracted from (a) whole plant (leaves, flowers and stems) (b) capitula (mg/10 g) (c) leaves (mg/1 0 g) and, (d) stems (mg/10 g) of D. odorata , collected at two sites for three geographical localities: Ingele, Ferncliffe and Dargle. . 127

Appendix A

Figure 1 Standard curve prepared from a hyoscyamine dilution series used to estimate the PA concentrations in aerial plant parts 162 XIV

List of Tables

Table 2.1 Taxa of exotic invertebrates and fungi released to control weeds of the Asteraceae. Taxa are ranked in decreasjng magnitude (based on the number of species released within the taxon) 25

Table 2.2 Weed species (Asteraceae) where good control has been achieved in at least one area of infestation and the agenUs which achieved control (Julien, 1992) 25

Table 2.3 Weed species (Asteraceae) partially controlled in at least one area of infestation and the agenUs which achieved this level of control (Julien, 1992) 26

Table 2.4 Taxa of exotic invertebrates and fungi used in the biocontrol of weeds of the Asteraceae. Number of species released within each order/family and their effect on weed populations is indicated. Families with agents which have achieved good control are highlighted in bold...... 27

Table 2.5 A list of the six most successful control agents released to control weeds of the Asteraceae (species which have achieved good to complete control on at least one weed in an area of infestation). The number of times these control agents have been released and the weed species they controlled is indicated (Julien, 1992) 29

Table 3.1 Location and altitude of populations of D. odorata used in this study

. · ·· · · 38

Table 3.2 Details of the pollen collected from the major pollinators of D. odorata: the number of species of each major group (family) , the number individuals, within each family, with D. odorata pollen and the average number of pollen grains per , is indicated. . 47 xv

Table 4.1 Average weight and size of achenes (±s.d.) of D. odorata 58

Table 4.2 Effect of depth of planting on seedling .emergence of achenes of D. odorata . . 62

Table 5.1 Mean PAR (umor's") [±s.d.] incident on transplants in each treatment ...... 73

Table 5.2 Summary table of the results (means (±s.d.); unless otherwise indicated) of measured characters of stolon spread for D. odorata plants grown at different light intensities for 12 weeks. n=60 plants measured for each character. 78

Table 6.1 Phytophagous associated with De/airea odorata. . . . . 89

Table 6.2 Life history data collected for P. rega/is. All measurements, except that of adult activity, based on data collected on galls initiated and raised to maturity in the laboratory (n-nurnber of galls) 101

Table 7.1 Pyrrolizidine alkaloid (PA) profile for De/airea odorata as determined by GC-MS (Stelljes et et., 1991). Type refers to the alkaloid properties (type of necine base) 115

Table 7.2 Mass spectra of pyrrolizidine alkalo ids extracted from De/airea odorata collected in Natal. 120

Table 7.3 Mass spectra of pyrrolizidine alkalo ids extracted from De/airea odorata by Stelljes et al. (1991). . 122

Table 7.4 Mass spectral data for 9-angelylretronecine (Stelljes et aI., 1991)...... 123 -1-

Chapter 1 General Introduction

1.1 Project description Delairea odorata Lem ., (syn. mikanioides Otto ex Walp.) commonly known as the Cape Ivy or German Ivy is a perennial semi-climbing plant that is indigenous to southern Africa. Here it occurs sporadically along the mist-belt region stretching from to the Eastern Cape, typically growing along forest margins or in the forest between 800 - 1900m above sea level. During the early 20th century D. odorata was introduced into several countries as a Victorian-era ground cover and house plant (Billiard, 1977; Jacobi and Warshauer, 1992; Archbald, 1995; Alvarez, 1997). Today, D. odorata is naturalised in many subtropical countries including the USA (California and Hawaii), Argentina (Buenos Aires), Chile, South Australia (New SouthWales and Victoria), Tasmania, Italy and northern Spain (Billiard, 1977; Fagg, 1989; Montenego eta/., 1991; Jacobi and Warshauer, 1992). In most ofthese areas Delairea is listed as an invasive species and a serious environmental weed warranting evaluation for biological control (Scott and Delfosse, 1992; Elliot, 1994; Archbald, 1995).

In exotic locations Delairea odorata is presently controlled by mechanical and chemical means (Cudney and Hodel, 1986; Archbald, 1995; Bossard and Benefield, 1995; Moore, 1997; Forbert, 1998). However, due to the high costs ofchemical and mechanical control coupled with the need for regular monitoring and follow-up treatments after removal, thousands ofdollars are currently spent on the control and eradication ofD. odorata (Robison, 1999). The only long term solution to the management ofD. odorata that seems practical is the introduction oforganisms that feed on it, i.e. biological control (Elliot, 1994; Balciunas pers. comm.). Ifsuccessful, biological control provides a low cost, long-lasting, alternative to mechanical and chemical control methods. Elliot (1994) suggested the release oftwo insects introduced to control Seneciojacobaeae L. onto D. odorata. During 1994, studies were conducted to test if these insects could reduce the vigour of D. odorata in the laboratory. No follow up report on these trials could be found. The alternative to using existing biocontrol agents is to search for the natural enemies ofD. odorata in southern Africa. This involves the development ofa classical biological control programme. Researchers from the USA are ofthe opinion that, in southern Africa, the growth ofD. odorata is stunted due to the action ofherbivores (Balciunas, pers. comm.) . -2-

This project was aimed at investigating aspects ofthe biology ofDelairea odorata in southern Africa. For a biological control programme to be effective knowledge ofthe distribution, biology and ecology of the plant species in its native environment is essential. It is hoped that this research will provide a series ofbaseline studies which will assist land managers and biocontrol specialists in developing control strategies for D. odorata.

1.2 Weeds: their definition, attributes and control One ofthe main threats to the biodiversity ofour environment is the direct destruction ofhabitats by the activities of man . Another serious, but largely underestimated problem, is the threat to natural and semi-natural habitats by the invasion of alien organisms (Rejmanek, 1995). Plant invaders pose a potentially lasting and pervasive threat as they may physically displace native plants and and in some cases change the soil composition enough to make native plant re-establishment impossible (Cronk and Fuller, 1995;Rejmanek, 1995). Conservationists are becoming increasingly aware ofthe problems posed by invasive plants as attempts to conserve native vegetation fail due to the disturbance effects created by vigorous aliens.

There is much general agreement as to which plants are weeds. However, definitions of what constitutes a "weed" or an "invader" and the terminology used in the study ofplant invasions and weed biology can be confusing (Pysek, 1995). Plants occurring in regions where they are not wanted have been termed weeds, aliens, invaders, exotics, immigrants, introduced, naturalised and colonisers (Baker, 1974; Heywood, 1989; Perrins et al., 1992; Pysek, 1995; Mack, 1996). In the literature a distinction between these terms is rarely made yet, technically, they do not reflect identical concepts.

The term weed is traditionally associated with plants invading highly disturbed man-made or agricultural habitats and thus often reflects an anthropocentric viewpoint (weeds are plants growing where they are not desired) (Baker, 1974; Cronk and Fuller, 1995; Rejmanek, 1995). Baker (1974) designated a plant a weed "if, in any specified geographical area, its populations grow entirely or predominately in situations markedly disturbed by man (without being deliberately cultivated plants)" .This definition does not include those plants which invade undisturbed and successionally advanced communities or natural/semi-natural habitats - the plants most frequently termed colonisers, invaders and aliens (plants spreading into areas where they are -3- not native) (Rejmanek, 1989; Pysek, 1995; Rejmanek, 1995). Furthermore, many non-native plants in protected areas are often classified as weeds ("ecological" or "environmental") because in national parks and similar areas non-native species interfere with the major management goals, i.e. protection ofthe native biota (Rejmanek, 1995). Since the term weed is so widely used, a more acceptable definition ofthis term might be:" A weed is a plant that is growing where it is not wanted" (Mortimer, 1990). Thus alien invaders may frequently be referred to as weeds .

An alien species has been defined as one which has reached an area as a consequence of the activities ofman or his domestic animals but is not necessarily a weed (pysek, 1995). It has also been termed an introduced species (Williamson and Fitter, 1996). An invader is an alien with increasing distribution in the wild, regardless of habitat (Pysek, 1995). Theoretically, an alien species can be considered invasive when it enters the exponential phase of spread (Rejmanek, 1995). An invasive species may also be referred to as a naturalised species (Williamson and Fitter, 1996). A weed may also be an indigenous species increasing in abundance and range. Prach and Wade (1992) termed a native species exhibiting such behaviour as expanding rather than invasive.

Over the past 30 years, numerous attempts to describe and define attributes common to plant invaders or weeds have been made. Contemporary scientists are asking the following questions: What makes a species invasive? Why are some plants more likely to become weeds than others? (Baker, 1974; Lawrence, 1980; Rejmanek, 1989; Rejmanek et al., 1991; Perrins et al., 1992; Vogt Anderson, 1995; Pysek, et al., 1995; Rejmanek, 1995; Mack, 1996; Williamson and Fitter, 1996; Vermeij, 1996). One ofthe first, and best known, attempts to describe attributes common to weedy species is Baker's (1965) list of"ideal weed" characters. In summary, he described an "ideal weed" as one which produces many widely dispersed, long-lived seeds which lack special requirements for germination, grows quickly, reproduces both vegetatively and sexually, flowers early, is self compatible and is a good competitor. Baker (1965) proposed a continuum of "weediness" where species with few "ideal characters" were unlikely to be weedy while those possessing many were likely to be invasive.

A review ofcontemporary literature indicates that many ofBaker's "ideal weed" characters are now being rejected. The two major reasons for this are: (1) many invaders have remarkably few ofthe Baker characters while other species, that have not become extensively naturalised, possess -4- many, e.g. Pinus pinaster Loud in South Africa (Kruger, 1977) (2) Baker's list is compiled for weeds which grow in areas markedly disturbed by man and while these characters may be common to agricultural weeds they do appear to hold for invasive plants of natural habitats (Rejmanek, 1989; Cronk and Fuller, 1995; Mack, 1996). However, despite a multitude ofmore recent attemp ts to define attributes common to environmental weeds (Rejmanek, 1989; Panetta and Mitchell, 1991; Perrins et al., 1992; Vogt Anderson, 1995; Rejmanek, 1995; Williamson and Fitter, 1996; Rejmanek and Richardson, 1996; Vermeij, 1996) as yet, few shared attributes have been found to be associated with invasive potential. The general conclusions from these studies are that there is no one suite ofcharacteristics which make a plant invasive therefore one species with a given set ofcharacters will be a weed in one place but not in another. Several predisposing factors have been suggested to increase the chance ofa plant becoming invasive. Some ofthese factors include taxonomic position, homoclines (climate), habitat, ecological status, dispersal characteristics, seed biology, genetic systems and mode ofreproduction (Groves, 1986). In other words, the potential for a plant to become a weed depends on: (1) the attributes ofthe species (life history, growth form and reproductive behaviour) (2) the environment in which it is growing (habitat and climate), and (3) the interaction between these two .

Weeds may be controlled by biological, chemical, mechanical or cultural means. Today, mechanical and cultural weed control are important only in the underdeveloped nations . These methods are labour intensive and frequently require rigorous monitoring and follow up treatments (Step hens, 1982). Following the discovery of potent herbicides during the 1940s, chemical control was seen as a permanent solution to weed problems. Soon, however, the ecological implications ofthese herbicides became apparent. This promoted the development ofintegrated control where pest infestations are reduced through a combination of methods - chemical, biological, mechanical and/or cultural. Only recently, however, has the concept of integrated control been applied with conviction to the control ofweeds. The development of herbicide­ resistant floras and their interaction with pests and disease organisms has made an integrated approach to weed control essential (Altmann and Campbell, 1977; Stephens, 1982).

While an integrated approach to weed control in agriculture is very effective, in the case of environmental weeds, which infest larger areas and where access is more difficult, biological control is frequently a better option (Bennett, 1986). Biological control is more convenient as -5- once an agent has established the control programme is self-perpetuating. Moreover, economic and social pressures to reduce reliance on pesticides is encouraging biologists to seek biological solutions to pest control (Stephens, 1982).

1.3 Biological control Biological control began in the thirteenth century when the ancient Chinese manipulated the predacious ant, Oecophylla smaragdina F. in mandarin orange trees to reduce the number of foliage feeding insects (DeBach, 1964). It was not until 1836, however, that the biological control ofweeds began. In 1836 the cochineal insect, Dactylopius ceylonicus Green, was taken from northern to southern India to control the cactus Opuntia vulgaris Miller (Goeden, 1988; Julien, 1989). Since then research in biological control ofweeds has dramatically increased and progress towards a greater understanding ofthe ecological principles underlying this practice has been made. Today, biocontrol is recognised as a multidisciplinary and applied science (Julien, 1989; Harris, 1991; Shepherd, 1993; van Driesche and Bellows, 1996).

What is biological control? A review ofthe literature reveals that the term "biological control" has no single definition. As argued by DeBach (1964) and Harris (1991), no single definition appears to be satisfactory because the term "biological control" is used with different meanings. Traditionally, biological control refers to the process whereby the natural enemies ofa plant (i.e. those that exist in nature), are used to control populations ofthe plant (DeBach, 1964; Anon, 1971; Dennil and Hokkanen, 1990; Shepherd, 1993; van Driesche and Bellows, 1996). Thus the term "classical biological control" refers to the process whereby the natural enemies of a weed are collected in the country oforigin ofthe weed and released in the country ofintroduction in order to control the weed. For the past 30 years the traditional approach to biocontrol has been much disputed and alternative methods to the" classical biocontrol" approach have been used (Dennil and Moran, 1989; Dennil and Hokkanen, 1990). A contemporary and suitable definition ofbiological control is thus "the study and utilisation ofparasites, predators and pathogens to regulate populations ofpests" (Hams, 1991).

Since the classical biological control approach is the method to be adopted for this investigation, the following section will focus on the principles and methods underlying the traditional approach. -6-

Classical biological control Biological control ofweeds is based upon a mutual dependance in the status of a weed and a phytophagous predator or fungal pathogen capable ofcontrolling it. It has usually been applied against alien weeds, the abundance of which is often due to their having escaped the natural enemies common to them in their native lands (Huffaker, 1964; van Driesche and Bellows, 1996). The main assumptions of the classical biological control approach are : (1) that the natural enemies (mono-or oligophages) are most likely to be host-specific and therefore safe for introduction into other countries and (2) that irrespective of the evolutionary age of the relationship between plant and natural enemy, and thus any possible homeostasis that may have developed between the two , it is the release ofthe biocontrol agents from their natural enemies that is the crucial factor in the achievement ofcontrol (Bartlett and van den Bosch, 1971; Dennil and Moran, 1989; van Driesche and Bellows, 1996).

A number ofscientists have challenged the basic assumptions ofclassical control (see review by Dennil and Moran, 1989) and today the search for effective biocontrol agents is less restrictive. The search for biological control agents can be extended to (1) areas other than the native home of a weed (2) to natural enemies which normally feed on a relative of the weed, and (3) to existing control agents (Bartlett and van den Bosch, 1971; Dennil and Moran, 1989; Dennil and Hokkanen, 1990). Despite the development of a number supplementary techniques many researchers still approach biological control in the traditional way (Dodd, 1961; Marohasy, 1989; Muller, 1989; Scott and Way, 1990; Gillet et al., 1991; McClay and Palmer, 1995).

Methods used in biocontrol studies Biological control is dependent upon the availability ofhost-specific control agents (Huffaker, 1964; Anon, 1971; Julien, 1989; Harris, 1991; Shepherd, 1993). Furthermore, the objective of biological control is not eradication, it is the gradual reduction ofa weed's density and impact (Huffaker, 1964; Julien, 1989). Agriculturalists have been reluctant to use biological control methods as the risks, pertaining to the introduction of new organisms into an already disturbed environment, are too great when compared with the chances ofobtaining an adequate degree of control. Added to this there is frequently conflict in the acceptance ofa given plant as a weed (Bartlett and van den Bosch, 1971). To reduce conflict and to increase confidence in biological weed control, extensive codes ofpractice and legislation are now in place to set out the required -7- procedures for the development ofa biological control programme (DeBach, 1964; Zwolfer and Harris, 1971; Harris, 1973; Wapshere, 1974; Lawton, 1984; Wapshere, 1989; Muller-Scharer et al., 1991; Shepherd, 1993; Cronk and Fuller, 1995;van Driesche and Bellows, 1996;FAO, 1997).

Today, classical biological control programmes proceed according to a general protocol designed to minimise the associated risks. This protocol was originally described by Zwolfer and Harris

(1971). Generally, the procedure is to: 1. Determine the weed 's suitability to biological control. 2. Conduct surveys ofthe weed's natural enemies. 3. Evaluate and select potential biological control agents from a range ofnatural enemies. 4. Conduct host-specificity studies of potential agents to ascertain the safety of their introduction into the new area. 5. Introduce the biocontrol agents into the control area. 6. Conduct evaluation studies .

The following section will expand on the biological and ecological theories upon which the above procedure is based.

Determine the weed's suitabilityfor biological control It is important to determine the suitability ofa weed for biological control before embarking on an extensive research programme. Thus, prior to the beginning ofthe actual search for biological control agents, the following basic information on a weed should be established : taxonomic position, biology, ecology, native geographic distribution, total present distribution, probable centre oforigin and that ofits close relatives and its economic importance (DeBach, 1964; Anon, 1971; Scott and Delfosse, 1992; Shepherd, 1993; van Driesche and Bellows, 1996).

A basic requirement of all projects is a thorough understanding of the ofthe target weed (Huffaker, 1964; Schroeder and Goeden, 1986; Harris, 1991; van Driesche and Bellows, 1996). A study of the taxonomy of the plant in the country of introduction as well as in the country oforigin ensures that agents are collected from the correct plant species. Projects have failed because of a lack of knowledge of the host plant; for example, early projects against Salvinia molesta D.S . Mitchell failed as this species had not been taxonomically separated from -8-

S. auriculata D.S. Mitchell. In absence ofknowledge ofS. molesta's native range, agents had been collected from S. auriculata and were thus unable to establish on S. molesta (Schroeder, 1985).

The distribution, pest status and economic impact ofthe weed in the countries ofintroduction should be ascertained through literature surveys to enable the researcher to find out what is known about the plant (Julien, 1989; McClay, 1989; Shepherd, 1993; van Driesche and Bellows, 1996). Given present economic conditions, effort should be focused on weeds which are serious economic problems and which offer a realistic chance ofsuccess. However, there may be little or no information on the economic impact ofa weed species, particularly ifthe weed in question is an environmental as opposed to agricultural weed, and in such cases the study may be limited to determining the biological suitability ofa weed as a target species (McClay, 1989).

A study ofthe biology and ecology ofthe plant should be made in both the country oforigin and introduction. This will help to determine why the plant is a weed, the vulnerable stages ofthe weed to nat ural enemies and whether there are any major differences in the life forms and/or habits ofthe plant. The study should determine ifthere are different ecotypes or biotypes ofthe plant present and should take into account seasonal variations in life history characters with respect to climate, temperature and rainfall (Andres et al., 1976; Shepherd, 1993; van Driesche and Bellows, 1996). Extensive intraspecific variation in the target weed can make biological control difficult as the ecotypes may differ in their susceptibility to the biocontrol agent (Andres et al., 1976; McClay, 1989; Shepherd, 1993). Furthermore, it may be difficult to identify effective biocontrol agents for a particular ecotype, and if several forms are present, several biotypes ofthe control agent may have to be introduced (McClay, 1989; Dennil and Hokkanen, 1990; van Driesche and Bellows, 1996). Critical to a biological control programme is determining the vulnerable stages in the plants' life history. Any agent which is to be introduced must attack the plant so as to limit its reproductive output, be that sexual or vegetative (MeClay, 1989; Shepherd, 1993; van Driesche and Bellows, 1996).

Natural enemies

Many types of herbivorous insects and pathogenic fungi have been considered as biocontrol agents (Crawley, 1989; Julien, 1989). The most severe constraint on the choice ofagents is their -9- host-specificity. Controversy exists over whether it is possibleto empirically analyse the qualities of candidate enemies and introduce only the best species (Huffaker et al., 1971; Hams, 1991; McEnvoy et al., 1993; van Driesche and Bellows, 1996). An introduced enemy's efficiency in controlling a given weed is inherently related to other stresses the weed faces from its direct competitors (Anon, 1971; Harris, 1991). Judgement of the "best" enemy in one vegetation complex will, therefore, have only limited meaning in another vegetation complex. However, ecological information and analysis ofprevious control attempts have been used to indicate which species should be tried first (Anon, 1971; Huffaker et al., 1971; Andres et al., 1976; van Driesche and Bellows, 1996). Effective biocontrol agents should possess the following attributes: 1. Adaptability to varying physical conditions. 2. Searching capacity, including general mobility (or capacity for dispersal). 3. Power ofincrease (in population size) relative to that ofits host. 4. Power ofhost consumption (i.e. negative impact on host growth capacity). 5. Other intrinsic properties such as synchronisation with the hosts' life history, host specificity and ability to survive host-free periods.

Evaluation ofnatural enemies should comprise the following stages (Anon, 1971): 1. Broad survey ofthe literature and regular sampling at local populations to determine the natural enemies. 2. Concentrated studies on single control agents that cover life history, distribution and parasitoids of the agent. 3. Host-specificity studies. Harris (1973) outlined some useful parameters for the selection of effective agents for the biological control ofweeds.

Host-specificity studies

One of the most important considerations in biological control is the host-specificity of the control agent as it is important to demonstrate that on introduction to a new area it will not damage plants ofeconomic and conservation importance (Huffaker, 1964; Zwolfer and Harris, 1971; Wapshere, 1974; Andres et aI., 1976; Wapshere, 1989; Blossey et al., 1994; van Driesche and Bellows, 1996). The absence ofinsects from plants related to the weedis a good indication ofhost-specificity as the propensity for feeding on other plants is usually greatest for those having -10-

taxonomic affinity with the host (Zwolfer and Harris, 1971; Shepherd, 1993). However, host recognition may depend on a single chemical or physical stimulus which has arisen independently in several plant taxa. Therefore, once suitable agents have been selected, scientific tests defining their host range need to be conducted (Anon, 1971; Wapshere, 1974; van Driesche and Bellows, 1996). These are in the form oflaboratory and field tests.

Laboratory tests Tests to determine the specificity of herbivorous arthropods have been conducted as either "choice" or "no choice" tests (Syrett, 1985; Adair and Scott, 1991; Dunn and Campobasso, 1993; Woodbum, 1993; Blossey et al., 1994; Wan et al., 1996). However, there has been considerable . variation in the experimental techniques used and critical plants tested (Harris and Zwolfer, 1968; Zwolfer and Harris, 1971;Wapshere 1974; Wapshere, 1989; Clement and Cristofaro, 1995). "No choice" tests are conducted by isolating the insect in a cage with a plant species to determine whether any feeding or oviposition can occur. In "choice" tests the agent is given a choice, in a single cage, of two or more plant species and feeding and reproduction on each is recorded (Zwolfer and Harris, 1971).

The criteria for selection ofplant species to test has been the subject ofmuch debate. Typically, the species selected are close taxonomic allies (same genus, tribe or family) (Harris and Zwolfer, 1968; Harris, 1973; Wapshere, 1974; Wapshere, 1989; Shepherd, 1993). The testing ofplants by phylogenetic methods is based on the assumption that related plants are morphologically and biochemically more similar than unrelated plants (Wapshere, 1974; Wapshere, 1989; van Driesche and Bellows, 1996). That closely related plants are morphologically and biochemically similar is a sine qua non since plant taxonomy and phylogenetic relationships have been based on these aspects. Tests should also be conducted on both cultivated plants botanically related to the weed and cultivated plants known to be attacked by organisms closely related to the biocontrol agent under investigation (Woodbum, 1993).

Field tests

Several authors have questioned the applicability ofresults obtained under laboratory conditions to the behaviour ofthe potential biocontrol agent in natural plant settings (Zwolfer and Harris, ·1971;Lawton 1984;Wapshere, 1989). Tests confining insects oftengive unnatural results because -11- dispersal is eliminated and this is an important part of the host selection process. Thus an organism may feed/oviposit on plants in the laboratory which it would not normally select in the field. The difficulties arising from laboratory testing can be avoided by testing for host-specificity under field conditions. In open field tests the selected arthropods can have "free choice" of plants without the constraints on their host selection behaviour (Maddox andSobhian, 1987; Clement and Cristofaro, 1995). Clement and Cristofaro (1995) reviewed the development and use ofopen field tests in host-specificity determination. They proposed that during insect screening programmes both laboratory and field methods should be used (Dunn and Campobasso, 1993; Blossey et al., 1994).

Introduction and evaluation ofbiocoritrol agents Once tests regarding the host-specificity ofa natural enemy are complete, decisions to introduce an agent into a weed-infested area can be made (Anon, 1971; van Driesche and Bellows, 1996). Such decisions are usually made at higher official levels than the research unit conducting the tests and regulations regarding the mode ofintroduction will vary depending on the organisation in charge ofbiological weed control for that specific area (country/state) (Shepherd, 1993). Once an enemy has been introduced, close descriptions ofthe destructive action (ifany) and patterns of destruction caused by the agent should be made if the results ofthe introductions are to be evaluated accurately (Shepherd, 1993; Cronk and Fuller, 1995; FAO, 1997). A principle tool in the evaluation ofthe impact ofintroduced agents is close studies ofthe changes in the general vegetation, in association with studies on the specific impact ofthe enemy on the weed in question (Anon, 1971; van Driesche and Bellows, 1996).

1.4 General background to the present study Presently chemical and mechanical methods are proving inadequate, and expensive, for the control of D. odorata. Consequently, biological control offers a seemingly attractive long term, economically viable alternative to the management ofthis plant. Despite the status ofD. odorata as an highly invasive species, very little is known about the biology and ecology ofthis plant in both its native and exotic environments. Ifa biological control programme for D. odorata is to be developed it is vital that aspects ofits biology be investigated. -12-

Little has been published about D. odorata in southern Africa, South America, Europe and Australia. Some work has been conducted into the status ofD. odorata in California and Hawaii (Cudney and Hodel, 1986; Jacobi and Warshauer, 1992; Chipping, 1993; Elliot, 1994; Archbald, 1995; Alvarez and Cushman, 1997; Archbald, 1997; Forbert, 1998) . However, most ofthis is unpublished research obtained largely from files of interested individuals and biological newsletters. Increasing concern about the rate ofspread ofD. odorata in California has recently motivated the development of several research initiatives aimed at investigating its biology and developing effective methods ofcontrol (Balciunas, pers. comm.; Robison, pers. comm.).

The following section is an account of the current status of knowledge on the taxonomy, distribution, biology and ecology ofD. odorata.

Taxonomic position Delairea odorata is a member ofthe Asteraceae tribe , subtribe Senecioninae (Jeffrey and Chen, 1984). The genus Delairea Lem. (syn: Senecio sect. scandentesDC.) consisting ofthe single South African species D. odorata is allied to the genera Mikaniopsis, Cissampelopsis and Austrosynotis (Jeffrey, 1986). All three genera are grouped into the Synotoid complex which is represented in Africa and Madagascar by scandent species. They are characterised by the presence of tailed anthers with sterile basal auricles (Jeffrey, 1986). Closely related species occurring in southern Africa include:Mikaniopsis cissampelina DC., Senecio cinarescens DC., SenecioquinquelobusDC. and Senecio tamoidesDC. (Arnold and de Wet, 1993; Jeffrey, 1986).

Delairea odorata is a bright green, slender, herbaceous twiner, stems and leaves slightly succulent, glabrous to glabrescent. Leaves alternate, deltoid-ovate, up to 70-80 mm in diameter, margin 6-10 lobed, base cordate. Petioles 70-80 mm, auriculate, often twisted. Heads discoid, many in congested compound corymbose panicles, terminating short lateral branchlets which are nude or with few reduced leaves. Involucre narrowly ca.mpanulate, 8, much shorter than the disc and keeled at the base. Flowers bright yellow, ovary 1.5 mm long, sparsely hispid (Hilliard, 1977). -13-

Distribution, status and community associations Delairea odorata is uncommon in South Africa. It occurs sporadically along forest margins in mistbelt vegetation stretching from the southern and Eastern Cape to Lesotho and grows as far north as Nkandla, Nhlazatshe Mt. and the Biggarsberg north ofLadysmith (Fig. 1.1). In South Africa it flowers in the late autumn and early winter months ofMay, June and July. Only one South African population seems to grow in an invasive smothering manner similar to plants where it has become an invasive weed. At Ingele Forest, Harding, D. odorata grows vigorously along the road and forest edges (pers. observation, Robison, pers. comm.).

In Australia D. odorata occurs in South Australia, Tasmania, New South Wales and Victoria. In Victoria it is now widespread in the southern districts from Portland and Warrnambool on the Western coast, to Wilsons Promonotory, and to Mallacoota in far east Gippsland (Fagg, 1989). D. odorata has been listed as an important weed of Australian conservation areas (Scott and Delfosse, 1992). It spreads most vigorously in moist semi-shaded environments such as damp gullies with tree cover and frequently occurs along roadsides (Fagg, 1989). It seems less successful in dry, exposed sunny conditions (Cox, 1998). Flowering occurs during the late winter early spring months ofJuly, August and September (Cox, 1998).

A map published by HEAR (1986) indicates that in Hawaii D. odorata is present on the islands ofHawaii, Maui and Oahu (Fig. 1.2). However, this reference gives no information on the habitat types in which D. odorata occurs. A study of the distribution ofD. odorata on the island of Hawaii was completed in conjunction with bird counts from 1976 to 1981 (Jacobi and Warshauer, 1992). This study found that on the island ofHawaii D. odorata occurs between 500 and 2500 m above sea level in areas with 1250 to 2500 mm annual rainfall. Major habitat types include xeric to mesic scrub and forest and hydric forest. Jacobi and Warshauer (1992) found most populations to be concentrated in the North Kona and Mauna Kea -Mauna Los saddle areas. In Hawaii D. odorata flowers in December and January.

Delairea odorata is also present in Europe. Almost nothing is known about the nature of the infestations in this region. Catalano et al., (1996) report its naturalisation in south and west Europe, including Italy. Researchers in northern Spain have also recently begun herbicide control tests (Robison, pers. comm.). The reported range ofD. odorata in North America extends from -14-

2523 25~4 25 R 2b?r6 12ff27 e8?g... 2'5·2'9 2530 2531 , 25;32 !---...-1 ~'"' T 7 i II'·~ (" _") I r -r V I I I I I I I I I 1 / ,I I ~M'< I _I V I 2923 29{A 2925 29f6 2921 29~8' 29,D9 aqo 29S1 I / J . L I r' w r / ,- i ' I ." I I I I I ! I I I I I I I I I I I , I I 11 , I ~BI 199,4 --.I 24 26 28 30 32 34 36

Fig. 1.1. Distribution ofDe/airea odorata in South Africa. (Distribution map is based on specimens lodged at herbaria: PRE and NU) -15- o

• Present DMay be present (search needed)

Fig. 1.2. Distribution ofDelairea odorata on the main Hawaiian islands as of1986 (HEAR, 1986). -16-

Oregon South through California into Baja California, though no comprehensive mapping efforts have yet been completed (Balciunas pers. comm.) . In California D. odorata has invaded the South Coast, the Central Coast and the North Coast, the San Francisco Bay area and many ofthe state's national parks (Anon, 1996; Alvarez, 1997). In the Golden Gate National Recreation Area (GGNRA) D. odorata is spreading more rapidly than any other non-native invasive plant species (Alvarez, 1997). Monitoring has shown that from 1987 to 1996 populations ofD. odorata moved from infesting less than 30 acres to more than 200 acres within the GGNRA. This represents an average population size increase of25 % per year (Fisher, 1997). The species has been rated as "highly invasive" by the new Jepson Manual and is also included in the California Exotic Pest Plant Councils' (CalEPPC) A-I list of "Most Invasive Wildland Pest Plants: Widespread". In California, D. odorata appears to prefer moist, shady environments along the coast. However, it is also spreading in riparian forest, coastal shrubland, grassland, Monterey pine forest, coastal bluffcommunities and seasonal wetlands. It has even been noted on serpentine soils (Archbald, 1995). Flowering occurs during the winter months ofDecember and January.

Basic biology In its exotic locations, D. odorata forms impenetrable mats in both the shade and the sun, climbs native shrubs and trees and often forms a dark canopy layer up to nine metres high, smothering native vegetation in the immediate vicinity (Fagg, 1989; Chipping, 1993; Alvarez, 1997; Fisher, 1997). D. odorata expands vegetatively as a vine through spreading stolons and frequently sends out runners in response to poor light conditions. These runners have leaves at each node and each of these nodes has the potential to become a new plant (Chipping, 1993). Growth rates of individual plants and populations measured at the GGt'-TRA have shown that increase in growth may be up to 30 cm per month per individual stem (Alvarez, 1997).

Delairea odorata has the ability to regenerate from small pieces ofabove ground stem material (Archbald, 1995) . Stem fragments as short as two cm carried by runoffor landscape machinery can take root and colonize new areas (Chipping, 1993; Archbald, 1995; Alvarez, 1997). In addition, it has thick rhizomes that support regrowth ofabove ground portions when shoots are damaged or removed. The system ofrhizomes was aptly described by Chipping (1993):" I could not understand how resprouting shoots kept showing up in areas unfit for growth down under the grass was a mass ofpurple, leafless , not particularly woody, and numerous enough to -17- look like the control panel wiring for a B-1 bomber". This strong predisposition for vegetative reproduction makes mechanical control ofthis species very difficult.

A series of experiments conducted by a student group in California have shown D. odorata to be drought tolerant. In a 10 week drought tolerance test they found that not a single well established plant was killed by moisture regimes ranging from 0-100 % for the full 10 weeks. Even cut stems left to dry on the laboratory table for the 10 week period were alive at the end of the test. Furthermore, when the drought stressed individuals were watered at the end of10 weeks they recovered normal water potential in their tissues after two days, even though the drought stressed plants were observed to have air embolisms in their tissues (Archbald, 1995).

Very little research has been conducted into the breeding system and seed biology ofD. odorata. In California, no viable seed has been collected and the plant appears to be spreading only by vegetative means (Balciunas, pers. comm.; Alvarez, 1997). In Australia and Hawaii D. odorata apparently spreads both vegetatively and by wind blown fruits (Jacobi and Warshauer, 1992; Cox, 1998). However, no data or research has been published in support ofthe above statement. Prior to this study no research had been conducted into the reproductive biology ofD. odorata in southern Africa. The nature ofthe contrasting modes of reproduction in exotic locations raises some interesting questions as to the genetic attributes and the nature ofthe spread of populations ofD. odorata in each ofthese areas.

Ecological effects ofDelairea odorata At the GGNRA, California, research on the impact ofD. odorata on coastal plant communities has shown that it is associated with significant reductions in species richness (Alvarez and Cushman, 1997). Grasses and annual species were found to be consistently missing from Delairea-infested plant communities. In addition, the abundance and richness ofseedling plants is significantly reduced indicating that after weed control there may be little re-establishment of native plants in the infested sites (Alvarez, 1997). There is also evidence that infestation is associated with a reduction oftwo insect orders (Coleoptera and Diptera) in two riparian plant communities which could affect other species densities dependent on these insects (Fisher, 1997). Furthermore, D. odorata is a threat at the species level, with four federally listed species and nine state listed species threatened by habitat modification, displacement and shading from this plant. -18-

Host and nectar plant populations for two endangered butterflies, the Mission Blue (Plebejus (Icaria) icarioides Boisduval (: Lycaenidae)) and the San Bruno Elfin (Callophyrs mossii Edwards (Lepidoptera: Lyacaenidae)) are also threatened by the spread ofD. odorata (Alvarez, 1997).

Delairea odorata contains pyrrolizidine alkaloids which are a group of 'highly toxic' secondary plant compounds. Pyrrolizidine alkaloids of a related species, Senecio jacobaeae, have been associated with growth depression, mortality and development ofhepatic lesions in rainbow trout (Hendricks et al., 1981). D. odorata has been associated with a toxic effect on aquatic organisms (Archbald, 1997). Preliminary studies, to determine the toxicity ofD. odorata, have shown that liquid extracted from blended plant material or small stem fragments has a significant toxic effect on the golden shiner (Notemigonus crysoleucus Mitchell), a common toxicological subject (Archbald, 1995). Further studies are being conducted with rainbow trout (Salmo gairdneri Richardson) and three-spine stickleback (Gasterosteus aculeatus L.), fish more characteristic of the coastal riparian habitats which D. odorata could invade. These results will indicate whether the threatened coho salmon (Oncorynchus kusutch Walbaum) and the California freshwater shrimp (Syncaria pacifica) will be directly affected by D. odorata infestations.

Mechanical and chemical control The environments in which D. odorata grows frequently makes access to the plant difficult and this hinders mechanical and chemical control (Moore, 1997). Effective mechanical control requires the removal ofall vegetation, native and exotic, from the infested area in order to gain visual and physical access to locations where stems emerge from the ground. Above-ground shoots are then pulled or cut away and the root mass dug out with hand tools (Archbald, 1995; Moore, 1997). At sites where the plant is growing mat-like on the ground it is sometimes possible to role up the entire infestation like a carpet (Archbald 1995). As D. odorata reproduces vegetatively it is difficult to dispose ofthe plant material. Usually, after clearing, the plant is removed from the site and laid out in the sun to dry to ensure complete desiccation. Continued monitoring and repulling at the cleared site is needed as resprouting occurs from remaining fragments (Moore, 1997). -19-

Preliminary herbicide trials, carried out by the Presidio National Park in San Francisco, have shown that spraying with Triclopyr", Glyphosatefand Sylgard" reduces resprouting in D. odorata. Furthermore, weed whipping and application of a low volume of'Triclopyr" successfully reduces above and below ground biomass (Archbald, 1995). Archbald (1995) observed that where plants were not weed-whipped prior to application ofthe herbicide, translocation ofthe herbicide along the rhizome was blocked by much thickened nodes or storage organs. Thus the rhizome on one side ofthe storage organ was killed, but the rhizome continuing from the other side was not visibly damaged. Other herbicides effective against D. odorata include Roundup" (active constituent: glyphosate) and Rodeo" (Archbald, 1995; Forbert, 1998). After spraying an infestation, regular follow-up treatments are essential.

1.5 Overall aims of the present study The aim ofthe present study was to survey the natural enemies associated with D. odorata and to determine their suitability to biological control. However, as indicated previously, prior to any extensive investigations on natural enemies and their effect on plant populations ofa pest, studies on the suitability ofa weed to biocontrol should be conducted. Critical to this is an understanding ofa weed's reproductive biology in both the native and exotic locations.

For this investigation emphasis was placed on:

1. Investigating the reproductive biology ofD. odorata in southern Africa. n. Identifying natural enemies which may be useful in the control ofD. odorata.

Finally, a study was conducted into the pyrrolizidine alkaloids ofD. odorata. The toxicity ofD. odorata to aquatic organisms has been associated with the presence ofpyrrolizidine alkaloids. Furthermore, several studies have already been conducted into the pyrrolizidine alkaloids ofD. odorata in the USA (Stelljes and Seiber, 1990; Stelljes et al., 1991). A study ofthe profile of these compounds for D. odorata across its native range will indicate to what extent the alkaloid profiles in native and exotic locations differ. Perhaps the breeding behaviour in exotic locations favours the formation of genetically isolated populations which may preserve their specific chemical characters. If so, this could have important implications for a biological control programme. -20-

Due to the diversity ofthis project, the theory, specific objectives, methods and results ofeach aspect ofthe research will be introduced and dealt with separately in the relevant chapters.

Chapter 1 introduces this study and also reviews the literature pertinent to biological control. Chapter 2 is a review on the biological control programmes targeted for weeds ofthe Asteraceae. Chapters 3,4 and 5 are investigations into sexual and vegetative reproduction in D. odorata. Chapter 6 details the natural enemies ofD. odorata and their potential as control agents. Chapter 7 is a study on the pyrrolizidine alkaloids ofD. odorata, and Chapter 8 is a summary ofthis study and outlines directions for future research. -21-

Chapter 2

Weeds of the Asteraceae and their biological control: a review

2.1 Introduction The Asteraceae as weeds If the evolutionary success of an organism is to be measured in terms of the numbers of individuals in existence, the extent oftheir reproductive output, the area ofthe world's surface they occupy, the range of habitats they can enter and their potentiality for putting their descendants in a position to continue the genetic line through time (Baker, 1974), then most members ofthe Asteraceae rate as highly successful. From an evolutionary point ofview the Asteraceae are regarded as one ofthe most advanced families and there are few families with such an abundance ofweedy members particularly in the temperate areas ofthe world (Heywood et al, 1977; Heywood,1989).

The Asteraceae is one ofthe largest families of flowering plants with 1100 currently accepted genera and 25 000 species. The diversity of this family is greatest in montane subtropical or tropical latitudes particularly in areas that border arid regions . Among the members of the Asteraceae are evergreen shrubs; annual, biennial and perennial herbs; tap-rooted or tuberous rooted perennials; trees, scramblers, climbers and succulents. The most characteristic feature of this group is the head-like , known as the capitulum, made up of numerous small individual flowers, the florets, surrounded by an involucre of protective bracts. This type of inflorescence is constant throughout the family and is sometimes accompanied by various modifications (Heywood, 1993).

Heywood (1989) analysed "A geographical atlas ofworld weeds" by Holm et al. (1979) and found that the families with the highest number ofweeds worldwide were the Asteraceae, with 224 genera and 830 species (13 %), and the , with 166 genera and 753 species (12 %), considered as weeds. Similar high percentages ofthese families have been recorded elsewhere in the literature, for example, in the alien flora ofindividual regions such as California, where 14 % belong to the Asteraceae and 15 % to the Poaceae (Rejmanek et al., 1991). The plant families containing the most weed species in South Africa, are the Asteraceae" Poaceae Fabaceae and -22-

Solanaceae (in orderofdecreasing magnitude) (Wells and Stirton, 1982). Among the families over represented in the Czech alien flora composition, the Asteraceae rate the highest, with 20.5 % present in the alien flora and only 10 % in the native flora (Pysek et aI., 1995). A review of "World Weeds" by Holm et al. (1997) again showed the same pattern, with the leading families being Poaceae (15 %) and Asteraceae (14 %) . The Poaceae and Asteraceae rank amongst the largest angiosperm families, perhaps it is the very features that have been responsible for their evolutionary success and adaptive diversity that are responsible for their successful spread and establishment as aliens worldwide (Heywood, 1989).

Attributes associatedwith the success ofthe Asteraceae The success of the Asteraceae has been attributed to their ability to tolerate a wide range of habitats, their high reproductive rate and their unique chemical compounds. The extreme plasticity ofthis family is reflected in their habitat accommodation where extreme xerophytes or mesophytes often occur in the same genus viz. Senecio. This ecological plasticity has been made possible through diversification ofgrowth habit, which includes everything from slow growing, cold enduring, hard woody shrubs to rapidly growing, cold-sensitive herbaceous perennials (Heywood et al, 1977). Their diversity ofgrowth habit has also contributed to their success as weeds and this is reflected in those species that are invasive - where annual, biennial and perennial herbs, shrubs and trees are all represented in the weedy flora of invaded regions (Heywood, 1989).

According to Heywood (1989) the factors which have contributed to the reproductive success of the Asteraceae are associated with the development of the capitulum, this includes: a geitonogamous breeding system (often superimposed with agamosperrny), increased pollinator attraction (associated with the aggregation of flowers) and the development of complicated dispersal mechanisms. The one-seeded fruit often has a pappus or wings to aid in wind dispersal. Hooks and spines frequently aid dispersal of seeds by animals and man (Heywood et al, 1977; Heywood, 1989).

The family is very distinctive in its chemical attributes. Although no single compound is unique to the family, the Asteraceae are remarkable in the range ofsecondary compounds present and also in the number ofcomplex structures known in anyone class (Heywood et al, 1977). Many -23- ofthe substances produced by the family are toxic or show other physiological activity and this provides a great deal ofprotection from phytophagous insects, overgrazing by livestock and use by man. It is for this reason that the chemical composition ofthe Asteraceae is believed to have

.played a part in their success as weeds.

Several members ofthe Asteraceae are serious noxious agricultural and/or environmental weeds in areas far removed from their indigenous ranges. They include: Chondrilla juncea L. (skeleton weed), Chromolaena odorata (L.) R. King & H. Robinson, Ageratina riparia (Regel) R. King and H. Robinson, Taraxacum species (dandelions), several Cirsium and Carduus species (thistles), and many Senecio's (Heywood, 1989). Since there are few families with such an abundance ofweedy members as the Asteraceae it is not surprising that a number ofspecies have been the targets ofbiological control programmes (Julien, 1989) .

Biological control ofthe Asteraceae The following section is a short review on the biological control programmes aimed at weeds of the Asteraceae. As the history of the biological control of weeds is characterized by both successes and failures (Crawley, 1989), many researchers are now analysing previous biological control programmes to determine whether patterns which emerge from such analyses will be useful in guiding future control attempts (Crawley, 1989; Harris, 1989; Julien, 1989; McEnvoy et al., 1993). Since this study is aimed at the biological control ofa species ofthe Asteraceae a review ofthe literature on biological programmes, ofweeds in this family, was carried out to determine any trends which may be ofassistance in the development ofa control programme for Delairea odorata.

2.2 The database and methods used for this review The following analysis has been based largely on Julien's (1992) catalogue ofweed biocontrol cases, specifically those ofthe Asteraceae. Use ofthis catalogue as a database limits the scope ofthis study. However, the catalogue is sufficiently comprehensive to reveal the general trends in biological control ofthis group. Where necessary data from other sources have been used. In an attempt to understand the functional causes ofthe successful programmes particular attention has been paid to biocontrol agents which have controlled their respective hosts. -24-

There is no absolute measure of"success" of a weed biocontrol project as each programme is designed for a specific weed problem. Therefore, whether a project is a success or not will depend on the aims of the weed controlling body (Crawley, 1989). For the purpose of this review, to categorise "success", biocontrol programmes have been divided into three categories: (1) those where the agent released did not establish or established but has had no effect on the weed population (2) those where release ofan agent has achieved partial control in at least one area of infestation, and (3) those where release ofan agent has lead to marked to complete control ofthe weed in at least one infested area. As the study is based largely on Julien's (1992) catalogue of biocontrol programmes, placement ofprojects into each ofthe above categories is partly limited by the information supplied in the catalogue.

2.3 Biocontrol agents used in the control of the Asteraceae Julien (1992) lists a total of117 weedy species for which exotic invertebrates and fungi have been released, ofthese, 32 species (27 %) belong to 18 genera within the Asteraceae. According to Julien (1992), the Asteraceae represent the largest group ofterrestrial weeds which have been targeted by biological control programmes. Second to the Asteraceae are the Cactaceae where 22 species (18 Opuntia spp.) belonging to five genera have been evaluated and subjected to biological control. Clearly the Asteraceae represent the largest and most diverse group ofweeds subjected to evaluation for biocontrol.

A total of81 species (77 invertebrates and 4 fungi) have been introduced into exotic locations to control weeds ofthe Asteraceae (Table 2.1). Invertebrates belonging to the orders Coleoptera, Diptera and Lepidoptera have been the most popular control agents followed by the pathogenic fungi (Table 2.1). The minor agents include Hemiptera, thrips, nematodes and acarines. Despite the number ofagents which have been used, only 12 weed species have been partially controlled by their agent in at least one region ofinfestation. Programmes against only four weed species have achieved good to complete control in at least one infested area (Tables 2.2 & 2.3). For 54 % ofweeds targeted for biological control, no control was achieved with the release of an evaluated biocontrol agent.

Table 2.4 illustrates the success ofthe invertebrate and fungi families which have been used in the biocontrol of weeds of the Asteraceae. Most of the exotic invertebrates which have been -25-

Tabl e 2..1 Taxa Of exotic invertebrates and fungi released to control weeds ofthe Asteraceae. Taxa are ranked in decreasing magnitude (based on the number of species released within the taxon).

Order No. spp (/81) % Total

Coleoptera 31 38

Diptera 23 28

Lepidoptera 19 23

Fungi 4 5

Thysanoptera 1 1

Hemiptera 1 1

Acarina 1 1

Nematoda 1 1

Table 2.2. Weed species (Asteraceae) where good control has been achieved in at least one area ofinfestation and the agent/s which achieved control (Julien, 1992) .

Weed species Agent species

Ageratina riparia Entyloma ageritinae Barreto and Evans (fungus: Ustaligenales) (Regel) R. King & H. Procecidochares alani Steyskal (Diptera: Tephritidae) Robinson Oidaematophorus beneficus Yano & Heppner (Lepidoptera: Pterophoridae)

Carduus nutans L. Rhinocyllus conicus Frolich (Coleoptera: Curculionidae)

Carduus thoermeri Rhinocyllus conicus (Coleoptera: Curculionidae) Weinmann Trichosirocalus horridus Panzer (Coleoptera: Curculionidae)

Senecio jacobaeae L. Longitarsusjlavicornis Step hens; S. jacobaeae Waterhouse & Longitarsus sp. (Coleoptera: Chrysomelidae) -26-

Table 2.3. Weed species (Asteraceae) partially controlled in at least one area ofinfestation and the agent/s which achieved this level ofcontrol (Julien, 1992) .

Weed species Agent species

Ageratina adenophora Procecidochares utilis Stone (Diptera: Tephritidae) (Sprengel) R. King & H. Robinson

Ambrosa artemisifolia L. Zygogramma suturalis Fabricius (Coleoptera: Chrysomelidae)

Baccharris halimifolia L. Trirhabda baccharidis Weber (Coleoptera: Chrysomelidae) mellyi Chevrolat (Coleoptera: Chrysomelidae), Oidaematophorus balanotes Meyrick (Lepidoptera: Pterophoridae)

Carduus acanthoides L. Rhinocyllus conicus (Coleoptera: Curculionidae) Trichosirocalus horridus (Coleoptera: Curculionidae)

Carduus pycnocephalus L. Rhinocyllus conicus (Coleoptera: Curculionidae)

Carduus tenuiflorus Curtis Puccinia carduorum Jacky (fungus: Uredinales) Rhinocyllus conicus (Coleoptera: Curculionidae)

Centaurea maculosa Lam. Urophora affinis Frauenfield (Diptera: Tephritidae) Urophora quadrifasciata Meigen (Diptera: Tephritidae)

Chondrillajuncea L. Puccinia chondrillina Bubak & Sydenham (fungus :Uredinales) Eiophyes chondrillae Canestrini (Acarina: Eriophyidae)

Chromolaena odorata L. Pareuchaetes pseudoinsulata Rego & Barros (Lepidoptera: Arctiidae)

Parthenium hystophorus L. Buccaltrixparthenica Bradley (Lepidoptera: Lyonetidae) Zygogramma bicolorata Pallister (Coloeptera: Chrysomelidae) Senecio jacobaeae L. Tyria jacobaeae L. (Lepidoptera: Arctiidae)

Xanthium strumarium L. Epiblema strenuana Walker (Lepidoptera: Tortrichidae) -27-

T able 2.4. Taxa ofexotic invertebrates and fungi us ed. in th e biocontrol ofweeds ofthe Asteraceae. Number of species released within each order/family and their effect on weed populations is indicated. Families with agents which have achieved good control a re highlighted in bold.

Good control Taxon Order/Family No effect Partial control

Coleoptera Anthribidae 1 Cerambycidae 3 Cbrysomelid ae 5 4 3 Curculionidae 7 2* Buprestidae Apionidae 1

Diptera Agromyzidae 2 Anthomyiidae 2 Cecidomyiidac 3 Syrphidae 1 Tepbritidae 13 3 1 Lepidoptera Arctiidae 2 Cochylidae 3 Gelechiidae 2 Geometridae 2 Lyonetiidae 1 Noctuidae Pterolonchidae 1 Pyralidae 1 1 Pterophoridae 1 1 Tortrichidae 1 1 Fungi Hyphomycetes 1 Ustilaginales 1 Uredinales 1 Hemiptera Delphacidae 1

Acarina Eriop hyidae 1

Nematoda Tylenchidae

Thysanoptera Phlaoethripidae 1

* The same two species have achieved partial or good control ofa weed in at least one locality. -28- released belong to the order Coleoptera with high success rates occurring in the Chrysomelidae and Curculionidae. Agents from these two families have achieved a measure ofcontrol against nine weed species (Tables 2.2 & 2.3). Amongst the dipterans, only the Tephritidae have been partly successful. The tephritid are favoured for their host-specificity and are frequently used as agents for control ofweeds ofthe Asteraceae. However, their success as control agents is debatable. Only four species (23 %) have achieved a measure of success against their targets. Within the Lepidoptera several have lead to some degree ofcontrol oftheir host, no particular family has been favoured. Out ofthe less frequently used groups (fungi, Hemiptera, Thysanoptera, Acarina and Nematoda), the fungi represent the most promising agents.

The following section will focus on each ofthe more successful groups, reference to previous effective biological control programmes is made.

Beetles In support ofthe results obtained in this review, Crawley (1989) found that the most successful weed biocontrol agents were beetles; particularly, those belonging to the families Curculionidae and Chrysomelidae. This is contrary to expectation as, generally, the endophagous herbivores, like the dipterans, would be the agents most likely to be successful; especially for the Asteraceae, where many tephritid flies are specific to genera within this group (Freidberg, 1984; Crawley, 1989; Harris, 1989). The biological attributes responsible for the success ofthe coleopterans as agents are not fully understood. Factors associated with high establishment probability may be involved, this includes good host seeking mechanisms, high rates ofpopulation increase relative to the host, long lived adults and high voltinism (polymorphism in invertebrates where some individuals enter diapause and some do not). Crawley's (1989) data suggested that size is an important factor for establishment. According to Crawley (1989) smaller insects are more likely to become established following introduction as biocontrol agents.

The Curculionidae are recognised as "premier biocontrol agents" (0' Brien, 1995; Syrett et al., 1996) . The most frequently released and successful biocontrol agent for the Asteraceae is Rhinocyllus conicus Frolich (Coleoptera: Curculionidae) (Table 2.5) (Julien, 1992). This thistle­ head weevil has been released against 10 weed species in eight countries (Argentina; Canada; New Zealand; USA; Australia; New Zealand and South Africa) on at least 21 different occasions -29-

Table 2.5. A list ofthe six most successful control agents (species which have achieved good to complete control on at least one weed in one area ofinfestation). The number oftimes these control agents have been released and the weed species they have controlled is indicated (Julien, 1992).

Species No. of Weed species major releases

Rhynocyllus conicus 21 Carduus acanthoides*; C. nutans**; C. (Coleoptera: Curculionidac) pycnocephalus *; C. tenuiflorus*; C. thoermeri**; Cirsium arvense; C. vulgare; Silybum marianum; Ageratina riparia

Trichosirocalus horridus 10 Carduus acanthoides*; C. nutans, C. (Coleoptera: CurcuIionidae) pycnocephalus; C. thoermeri**; Cirsium palustre; Cirsium vulgare

Longitarsusjacobaeae 2 S.jacobaeae** (Coleoptera: ChrysomeIidae)

Procecidochares alani 2 Ageratina riparia** (Diptera: Tephritidae)

Oidaematophorus benejicus 1 Ageratina riparia** (Lepidoptera: Pterophoridae)

Entyloma ageratinae 2 Ageratina riparia ** (Fungus : UstilaginaIes)

* Indicates partial control attained in at least one area ofweed infestation. ** Indicates good/complete control attained in at least one area ofweed infestation. -30-

(Julien, 1992). The weevil has achieved some measure of control on five species of thistles belonging to the genus Carduus. The adults feed on stems, leaves and bracts offlower heads and also oviposit on these bracts. Most control resides with the larvae which feed within the receptacles inhibiting seed production (Surles and Kok, 1978; Dunn and Campobasso, 1993). Carduus nutans L., Carduus pycnocephalus L., Carduus tenuiflorus Curtis and Carduus thoermeri Weinmann are all annual or biennial herbs propagated principally by wind dispersed seeds. Destruction ofthe thistle heads by the larvae thus reduces seed output and hence achieves a measure of control (Holm et a!., 1997). Trichosirocalus horridus Panzer (Coleoptera: Curculionidae), a curculionod, has also achieved a measure ofcontrol on several thistle species (Table 2.5). The larvae ofthis damage thistle rosettes by feeding on meristem tissue. In some instances the most effective control has been achieved when both agents have been targeted against the same weed (thistle species) (Dunn and Campobasso, 1993).

More species ofChrysomelidae have been used as weed biocontrol agents than any other family ofinsects (Syrett et al., 1996). All species are phytophagous, they are generally host-specific and often cause complete defoliation oftheir host. Furthermore, they are usually easy to rear, fecund, disperse well and reach high densities in field populations (Syrett et al., 1996). Longitarsus jacobaeae Waterhouse, a chrysomelid, has achieved successful control ofSeneciojacobaeae L., a serious weed, in a number oflocations (Table 2.5) (McEnvoy, 1984; James et a!., 1992; Julien, 1992; McEnvoy et a!., 1993). This plant is a biennial or short-lived perennial reproducing principally by wind dispersed seeds but is also able to reproduce vegetatively by root and crown buds (James et al., 1992; McEnvoy et a!., 1993). The first agent to be released against S. jacobaeae (tansy ragwort) was Tyriajacobaeae L. (Lepidoptera: Arctiidae). The larvae ofthis feed on the leaves ofthe ragwort, and although they may completely defoliate the plants, their activity has been shown to increase plant density (Dempster, 1971). Later the ragwort flea beetle, Longitarsus jacobaeae (Coleoptera: Chrysomelidae), was introduced. It is by far a superior control agent. The adults ofthe flea beetles feed on leaves. Larvae damage the plant by boring into the roots and petioles where they feed throughout the winter (lames et a!., 1992). Experiments have shown that the two insects together have a greater impact on the plant than either insect acting alone (lames et al., 1992). However, high mortality ofyoung plants caused by the boring larvae of the flea beetle indicate that the beetle alone can be an effective agent (lames et al., 1992; McEnvoy et al., 1993). The ragwort flea beetle is said to epitomise the -31-

"search and destroy" strategy, in which the agent is monophagous on the host and highly capable offinding and destroying it.

Flies The family from which the most species have been released, for the control ofthe Asteraceae, is the Tephritidae (Table 2.4). These flies have been only partly successful control agents (Syrett et al., 1996). Seventeen species of tephritids are listed in Julien's (1992) catalogue of weed biocontrol agents for the Asteraceae. Only three out ofthe 17 species have partially controlled their target weed and only one species has inflicted substantial damage to its host (Tables 2.3,2.4 & 2.5). The tephritids have been selected as biocontrol agents largely for their host- specificity. A tephritid genus is usually restricted to a specific group ofplants, usually at the level of tribe or below (Edwards et al., 1996).

Procecidochares alani Steyskal (Diptera: Tephritidae) and Procecidochares utilis Stone (Diptera: Tephritidae), both stem gallers, are credited with the control ofAgeratina riparia (syn: Eupatorium riparium) and Ageratina adenophora (Sprengel) R. King and H. Robinson (syn: Eupatorium adenophorum) respectively and this represents the best result to date (Tables 2.2, 2.3 & 2.5) (Bess and Haramoto, 1959; Dodd, 1961; Harris, 1989; Julien,1992; Morin et al., 1997) . Both of these plants are serious economic weeds in many countries. Ageratina adenophora is a perennial which depends on both sexual and vegetative reproduction for population increase. Procecidochares utilis eggs are deposited between the young paired leaves before they unfold, where they hatch and feed, producing a gall. Control is manifested in shortened stems and reduced foliage production (Bess and Haramoto, 1959) .During a dry season fewer young leaves are produced by the weed and hence there are fewer ovipositional sites. Continued heavy attack by the fly causes the growing tips to die, reducing population spread.

Several tephritid species which form flower head galls have also been employed for biological control of the Asteraceae (Freidberg, 1984; Hams, 1989; Julien, 1992) . Seven species of Urophora, a flower-head galler, have been released against thistles and knapweeds (Julien, 1992). As most ofthese weeds are herbaceous annuals, biennials or short lived perennials propagated mainly by wind dispersed seeds, destruction ofthe seed head has a bearing on the reproductive capacity ofthese plants. The flower-head gall acts as a powerfull metabolic sink and therefore -32-

accounts for much energy loss from the capitulae ofthe host (Hams, 1989). Although many species ofUrophora have established on their respective hosts, Urophora affinisFrauenfeld and Urophora quadrifasciata Meigen are the only two species to have partly controlled their target weed, Centaurea maculosa Lam. (Table 2.3 ) (Julien, 1992).

The high rate of failure of many tephritid species to achieve control of their hosts is partly attributed to parasitism oftephritids in the galls (Edwards et al., 1996). Parasitism of tephritid galls by parasitic wasps is not uncommon and may effectively reduce agent population numbers and hence their control potential (Crawley, 1989). Unsuitability ofthe target weed as a host may also account for establishment failure, i.e. the chosen biocontrol agent cannot breed successfully on the target weed (Crawley, 1989; Harris, 1989). Furthermore, an agent can only be successful if it targets the life strategy ofthe plant. A tephritid fly which reduces seed production will have no effect on the plant population if reproduction is not limited by seed.

Moths Out ofthe 19 species ofLepidoptera released onto the Asteraceae, only one has achieved good control of its host (Table 2.4) (Julien, 1992). Oidaematophorus beneficus Yano and Heppner (Lepidoptera:Pterophoridae), a plume moth, has achieved good control ofA. riparia, commonly known as the mist flower, in Hawaii (Morin et al., 1997). On hatching, the larvae of0. beneficus feed in situ on the underside ofthe leaves. Later, they move to the new leaves at the terminals of shoots where they consume all parts of the young leaves. In Hawaii, the plume moth established rapidly on A. riparia. It is most active in August which coincides with the period of maximum growth and can completely defoliate the plants (Morin et al., 1997). However, at low altitudes and in the warmer regions, egg parasitism by braconid wasps reduce the impact ofthis species (Yano and Heppner, 1983; Morin et al., 1997).

Pareuchaetes pseudoinsulata Rego Barros (Lepidoptera: Arctiidae) has partly controlled C. odorata in Ghana, Guam and parts of Asia (Julien, 1992; Timbilla and Braimha, 1996). The damage caused to the terminal and axillary buds ofC. odorata by the larvae ofP. pseudoinsulata reduces seed set and consequently has the potential to reduce populations ofthis weed. Various degrees ofcontrol ofC. odorata have been achieved using this agent (Julien, 1992) . Timbilla and Braimha (1996) are ofthe opinion that effective control ofthis weed in most locations will require more than just defoliation by P. pseudoinsulata. -33-

Crawley (1989), found that the Lepidopterarated highly amongst the "poor establishers" for weed biocontrol agents and attributed this to high levels ofpredation. Despite the above examples, the Lepidoptera are not outstanding as control agents for weeds of the Asteraceae (Table 2.4). Crawley (1989) suggested that as the Lepidoptera are generally external feeders, their eggs and larvae are extremely vulnerable to generalist predators and, coupled with this, they are especially prone to viral and fungal diseases (Entwistle 1983; Crawley, 1989). No doubt these factors reduce the probability that an introduced agent will establish.

Pathogens The use ofplant pathogens as biological control agents has increased dramatically sincethe 1970s (Freeman and Charudattan, 1984; Julien, 1989). Such research holds much potential as fungal pathogens generally have host ranges limited to one or a few species and are therefore ideal biocontrol agents (Watson, 1984). Two ofthe four species listedas control agents in Julien's (1992) catalogue have achieved a measure ofsuccess against their target (Table 2.4). Entyloma ageratinae Barreto and Evans (syn. Cercosporella ageritinae, Entyloma compositarum) causes leafblight on A. riparia and has achieved striking control ofthis weed in Hawaii (Trujillo, 1984; Julien, 1992; Morin et a!., 1997). In 1975, inoculation of the plants with the fungus caused population decreases ofup 80 %, moreover, at several sites this was achieved injust nine months (Trujillo, 1984). The rust fungusPuccinia chondrillina Bubak and Sydenham killsthe leaves and I stems ofChondrillajuncea and subsequently reduces flowering. The specificity ofthis pathogen is so strict that certain strains are virulent against only one form of C. juncea.

2.4 Summary and conclusions Two important points emerge form this review: firstly, regardless oftype, a good agent is one which targets the life strategy of the host plant; secondly, good control of a weed is often achieved when multiple species introductions are made. Even if an agent is host-specific and has the ability to defoliate its host, if plant reproduction (vegetative or sexual) is not limited by herbivory then control is not initiated. Successful agents are also those not prone to attack by parasitoids in the new environment. Tephritids are potentially ideal candidates for biocontrol of the Asteraceae - they are host-specific (largely on the Asteraceae) and also endophagous herbivores. However, parasitisation of the flies by native parasitoids of the new environment often limits their establishment on their host plant. This analysis has also shown that trends in the -34- biological control ofthe Asteraceae echo that ofweed biocontrol in general (Crawley, 1989). Coleopterans are usually the most successful agents, and this is attributed to their host-specificity, ease ofdispersal and powers of consumption (Crawley, 1989).

Although this analysis is useful in highlighting groups which may contain potential control agents for D. odorata, it would be a mistake to read too much into the results. Reasons for failure ofa biocontrol project are often many, complex and interacting; for example, genetic variation ofthe plant population often means only a fraction of the weed is susceptible to control by an introduced agent. Thus potential agents should not be rejected merely because they have a poor "track record". Rather the results ofthis review should be used to highlight the weaknesses of a potential agent, thorough investigation into these aspects prior to release may help to determine whether successful establishment and control are feasable.

Julien (1989) highlighted the use of pathogens as control agents. As of 1989,66.6 % of the pathogens released had achieved control oftheir target weeds. Use ofpathogens has generally been restricted due to concerns about their stability and specificity (Julien, 1989). However, these problems are being overcome by the development of guidelines for testing (Freeman and Charudatten, 1984, Julien, 1989). Perhaps the best option for control ofD. odorata is by the introduction ofa host-specific microorganism or the development ofa bioherbicide. -35-

Chapter 3

Floral biology, breeding system and pollination

3.1 Introduction The search for correlations between colonising ability and breeding system has been a recurring theme in invasion biology (Baker, 1955; Baker, 1965; Price and Jain, 1981; Carr and Powell, 1986; Abbot and Forbes, 1993; Rejmanek and Richardson, 1996; Williamson and Fitter, 1996). It was little more than 40 years ago that Baker (1955) pointed out the correlation between the breeding system ofan organism and its ability to establish sexual populations after a long distance dispersal event. Baker (1965) suggested that for the "ideal weed" the most suitable breeding system for colonising new areas is "self compatible", but not completely autogamous or apomictic . Self-compatibility allows seed set at low population density in the absence ofsuitable pollinators. However, obligate self-pollination can lead to inbreeding and associated lack of variation and vigour in the population. For many years it was generally assumed that only self­ compatible species have weedy potential and although there is evidence in support ofthis, more recent studies have shown that the type ofbreeding system may not always be an important factor in determining invasiveness (Brown and Marshall, 1980; Price and Jain, 1981; Carr and Powell, 1986; Noble, 1989; Perrins et aI., 1992; Cronk and Fuller, 1995; Mack, 1996; Rejmanek and Richardson, 1996; Williamson and Fitter, 1996).

Although the breeding system of a plant may not be a determinant ofits invasive potential, it is an important factor to consider in the development ofa biological control programme. This is because the genetic structure ofpopulations of a weed influence the ability ofintroduced pests or pathogens to substantially reduce host stand density. In their analysis on biological control and the reproductive mode of weeds, Burdon and Marshall (1981) found a preponderance of apomictic species selected as targets in biological control programmes. Furthermore, the degree ofsuccess achieved in weed control was significantly affected by the reproductive mode ofthe target species - those reproducing asexually being more successful. The greater efficacy of biological control agents against asexual weeds arises because oftheir tendency towards genetic homogeneity, thus one biotype ofbiocontrol agent can potentially target all populations ofthe pest plant (Burdon and Marshall, 1981). -36-

Plant populations which arise from small inputs, as is often the case with colonising species, almost invariably show considerably less genetic variation than those found within naturally occurring populations ofthe same species (Harper, 1977; Burdon and Marshall, 1981; Cronk and Fuller, 1995). For agamous and clonally reproducing plants the very limited nature ofthe original introduction may result in weed populations which are genetically uniform. In sexually reproducing species, although the establishment process clearly reduces the variability of the population, recombination ofgenes from different parents results in stands containing a range of different genotypes (Burdon and Marshall, 1981). A lack of quantitative knowledge on the population genetics ofa weedy species precludes any attempt to pursue the question ofits genetic diversity and thus potential susceptibility to a biocontrol agent. However, a knowledge ofthe breeding system ofa plant species may provide some indication on its population structure. As a first approximation, it seems reasonable to assume that apomictic species will be far less variable on average than species which undergo sexual recombination every generation.

The present study The investigation on the breeding system ofa weed is not only important in terms ofcontributing towards an understanding of its population structure but also vital to the choice of effective biocontrol agents. As no literature has been published on the floral biology, breeding system and pollination ecology ofD. odorata no predictions regarding the genetic structure ofpopulations and hence the suitability ofD. odorata to biocontrol can be made. This study attempts to address some ofthese aspects for D. odorata in southern Africa. The specific question to be addressed by this study pertains to the breeding system ofD. odorata: Is D. odorata self-compatible or an obligate outbreeder? To complement the study on the breeding system, aspects of the floral biology and pollinator ecology ofD. odorata were investigated.

Experiments were also carried out to determine whether seed set in D. odorata is pollen limited. In addressing the question ofpollen limitation it is important to understand the breeding system of the plant. If plants are capable of producing seeds through autogamy or self-pollination, a reduction in pollen receipt may not lead to a decline in the number ofseeds produced, although average seed quality may change considerably(Vaughton, 1988). Alternatively, self-incompatible species require foreign pollen for seed set. -37-

Delairea odorata expands vegetatively as a vine through the spread ofstolons and frequently sends out runners which have the potential to root at each node, each developing into a new plant. With this clonal growth habit there is the potential for stands ofD. odorata to consist ofonly one or a few genotypes (Chipping, 1993). Such clonal growth is widespread among plants and clones have the potential to reach enormous sizes ego Eichhornia crassipes (Martius) Solms­ Laubach (Widen and Widen, 1990; Cronk: and Fuller, 1995). It follows that in clonal plants as the size ofthe clone increases flowers will become increasingly surrounded by flowers ofthe same individual (Hand el, 1985; Eriksson and Bremer, 1993). Observations have shown that most movements by pollinators in natural populations are between very near neighbours and therefore with increased clone size, geitonogamous pollination (pollination between different flowers of the same clone) increases (Widen and Widen, 1990). Furthermore, restricted gene flow favours the development oflocal structure in which neighbours on average are more similar genetically than are individuals separated by larger distances (Levin and Kerster, 1974; Arroyo, 1976; Hessing, 1988; Waser and Price, 1989). A negative correlation between fecundity per flower and plant size, attributed to insufficient out-pollination, has been reported for self-incompatible species (Carpenter, 1976; Andersson, 1988; Eriksson and Bremer, 1993). Insufficient pollination could therefore be a problem in self-incompatible species with extensive clonal growth. Furthermore, self-incompatible species are often insect pollinated and restrictions in pollen dispersal could therefore play a part in determining their fecundity, since they must outcross (Widen and Widen, 1990). This investigation aimed to determine the natural seed set ofD. odorata in its native environment and to compare this with seed-set of hand-pollinated flowers .

It is important to note that pollen limitation of the number of seeds or fruits may occur in a number ofother ways; and these must be considered in any study on pollen limitation (Goldingay and Whelan, 1990). Firstly, there may be low numbers of pollinators (Whelan and Goldingay, 1986). Secondly, there may be a suitable number of pollinators ordinarily but, due to profuse flowering at the time ofthe study, the pollinators become satiated (Copland and Whelan, 1989). Thirdly, pollinator visits may result in ineffective pollen transfer (Whelan and Goldingay, 1989). -38-

3.2 Materials and methods Study sites To detect possible differences in the breeding system of this species, populations from three locations across KwaZulu-Natal (Harding, Dargle and Pietermaritzburg) were included in the study. Table 3.1 lists the specific location, altitude and coordinates ofeach study site. All three study sites fall into the mist-belt region characterised by cool, tall forests usually dominated by Podocarpus species (Pooley, 1993).

Table 3.1. Location and altitude ofpopulations ofD. odorata used in this study.

Locality Site Altitude (m) Co-ordinates

Harding Ingele Forest 1300 30032'S, 29°43'E

Dargle Maritzdaal 1200 29°32'S,30001'E

Pietermaritzburg Femcliffe Nature Reserve 800 29°37'S, 30020'E

Over 70 % ofthe annual precipitation at the three study sites is concentrated between November and March, with a dry season from May to August. Rainfall is highest during the summer

0 (November to February) when the average monthly temperature ranges between 19 0 C and 22 C. Mean annual rainfall for all sites is above 800 mm, with a maximum of 1031 mm recorded for Ingele Forest. Dargle is the coldestlocation, average minimum temperatures range from 0-5 °C and frosts are frequent during June and July. Average minimum temperatures for Ingele and Femcliffe are above 5 °C and frosts are less frequent (CCWR, 1999).

Floral biology Observations on the development offlowers ofD. odorata were made both in the field and in the laboratory. Flowering phenology, at the population level, was monitored at each study site. At the start of the breeding season 50 immature inflorescences were tagged and the number of capitula in each recorded. For the rest ofthe season, weekly records on the development ofeach capitulum in each synflorescence was made. For this study, four stages in the development of the capitulum were recognised: (1) young bud (all florets closed) (2) opening (some florets opening) (3) full bloom (all florets open), and (4) withering! fruit set. Each week, for each tagged inflorescence, the number of capitula in each of these stages was recorded. Flowers damaged by herbivory or fungi were recorded in a separate category. -39-

Experimentalpollinations All experimental pollination was done by hand in the field. Each capitulum used in experimental pollinations was protected from contamination by a fine (0.5 mm) nylon-mesh pollination bag. . As the inflorescence is determinate with the central or apical capitulum maturing first, bags were frequently visited to ensure pollination occurred when the stigmas were receptive; judged by the spreading ofthe stigmatic lobes.

The breeding system was determined by a series offive treatments ofthe capitula: 1. Control. The capitula were tagged but otherwise unmanipulated to test for natural seed set under field conditions. 2. Cross-pollination. Immature capitula were bagged, as the stigmas became receptive they were pollinated with pollen from flowers collected at least lam away, and rebagged until fruit set. Pollen from each donor was applied by rubbing a whole capitulum ofthe donor plant onto the recipient capitulum, a technique similar to that used by Carr and Powell (1986) and Sobrevila (1989). As floret maturation within the capitulum is acropetalcross­ pollination was carried out several times to maximise fruit set. 3. Self-pollination. Immature capitula were bagged, as the stigmas became receptive they were brushed with pollen from the same head and rebagged until fruit set. 4. Apomixis. Immature capitula were bagged, as the stigmas became receptive they were removed with a sharp razor-blade, and the flowers rebagged until fruit set (emasculation was not possible as the florets are very small). 5. Pollinator exclusion. The capitula were not bagged or manipulated in any manner. Plants were kept in a greenhouse out ofcontact with pollinators. Fruit set was determined at the end ofthe season.

Fruit set was determined in the laboratory with the aid ofa dissecting microscope. Fertile fruits are plump and easily recognised, a pair of fine forceps was used to squeeze these achenes to confirm the presence of an embryo. Developed seeds filled the achene, while seedless achenes were characterised by the shriveled remains ofthe ovary. -40-

Pollinators Foragers on the inflorescences ofD. odorata were netted and killed in jars charged with ethyl acetate. Fieldworkto collect pollinators was carried out at only two ofthe study sites, Dargle and Ferncliffe (1998 and 1999) due to logistical constraints (Harding is over 200 km away from Natal University) . All pollinators were collected between 9:00 am and 4:00 pm. Following collection, the specimens were pinned, examined for the presence of pollen and kept in insect boxes for subsequent identification. To verify the presence ofD. odorata pollen, a small piece offuchsin jelly was rubbed over the body ofeach pollinator (to remove any pollen grains), placed on clean glass slide, heated and covered with a coverslip. This makes a permanent slide which can be used for qualitative and quantitative observations ofpollen loads.

Pollen limitation To test for pollen limitation of seed set in D. odorata, hand pollinated infructescences were compared to those which received open (natural) pollination. Zimmerman and Pyke (1988) outlined protocols for pollen limitation experiments. The tests used in this study are based on the methods they suggested but were adapted to suit the conditions of this study. Hand pollinations (HP) were performed as described in the previous section (Treatment 2: cross­ pollination). Two controls were used inthe investigation. On ramets with capitula which received artificial (hand) pollination (i.e. experimental plants) there were control capitula (EC) . These were open pollinated capitula, adjacent to the hand-pollinated ones, tagged at the start of the flowering season and left unmanipulated . According to Zimmerman and Pyke (1988) , if these capitula have the same seed set as those receiving an extra pollen load then seed set may be limited by resources other than pollen. However, ifcapitula receiving additional pollen show a higher seed number than control capitula the situation is ambiguous. Enhanced seed set in some capitula may occur at the expense ofreduced seed set in other capitula on the same plant and the observed pattern may simply be the result ofallocation ofresources to particular flowers. It is for this reason that the second control was necessary. These were open pollinated flowers on ramets which did not have any hand pollinated flowers (i.e. control plants) (CC). Ifcontrol capitula on both control and experimental plants show the same achene number then capitula receiving extra pollen have set more seeds at no expense to the other blossoms open at the same time.

Tests for pollen limitation of seed set were carried out on populations of D. odorata, at -41-

Maritzdaal (Dargle) and Ferncliffe Nature Reserve, during 1999. At the end ofthe season fruit set for each treatment was determined by counting the number offruits set per capitulum.

Data analysis Kruskal Wallis ANOVA was used to test the effect ofthe pollination treatments on seed set. The statistical package used was Minitab? (1995).

3.3 Results Floral Biology Capitula ofD. odorata are discoid, bisexual and protandrous and posses a standard number of10 florets per capitulum. The florets are tubulate, posses a five-lobed corolla and have a uniloculate inferior ovary. The synflorescence is determinate with the central capitula maturing first as opposed to the sequence of development offlorets within the capitulum where development is acropetal, the marginal florets developing first.

Floral development is very similar to its nearest ally Senecio L. (Lawrence, 1985) and typical of the Asteraceae. In the family the five anthers are syngenecious forming a collar around the style (Fig. 3.1) . The anthers dehisce introrsely while the florets are still closed but selfpollination at this stage is avoided as the receptive stigmatic surfaces are closely adpressed (Fig. 3.2). At this stage the style apex is positioned at the base of the anthers. The growing style then passes through the anther collar and brushes the pollen out ofthe anther cylinder by means ofhairs on its outer surface (Fig. 3.3). The pollen mass extruded in this manner can then be removed by pollinators. Thus pollination can only occur after the style branches have extended beyond the collar ofanthers, at which point the branches spread apart and expose the stigmatic surface (Fig. 3.4).

Flowering phenology was monitored at Ferncliffe, Dargle and Ingele (Fig. 3.5 a-c). At all three sites flowering commenced in early April (buds) . This coincides with the first drop in temperature and rainfall at all sites. Peak flowering period (full bloom) occurred over the interval oflate April to late May. Thereafter most ofthe flowers started to wither and set seed. By mid-July most of the fruits had been dispersed by wind. Approximately 18 % offlowers at each site were damaged by fungi and/or herbivory (Fig. 3.5 a-c). The breeding season thus covers approximately three months (April, May, June), through the autumn to early winter period. T -

-42-

Fig. 3.1. Intact anther collar, ofaD. odorata floret, consisting offive syngenesious anthers.

Fig. 3.2. Status ofthe style when the D. odorata floret is still closed. Floret and anther collar have been prised open to show the style apex positioned at the base of the anther collar. -43-

Fig. 3.3. The style pushes through the narrow anther collar of the D. odorata florets and removes pollen out of the cylinder by means ofhairs on the outer surface.

Fig. 3.4. After the style has extended beyond the anther collar, the style arms spread apart and expose the stigmatic surface. a -44- 100

80

60

40

20

0 23/04/ 29/04/ 7/04/ 12105/ 22105/ 27/05/ 4/06/ 11/06/ 25/06/

b 100

80

.- 60 ~ Cl '-' -~ 40 .....-= c.. 20 ~ CJ

0 10/04/ 17/04/ 3/05/ 15/05/ 2106/

C 100

80

60

40

20

0 7/041 16/04/ 22104/ 28/04/ 06/05/ 13/05/ 22105/ 26/05/ 01/06/

date

Fig. 3.5 a-c. Phenology of D. odorata at (a) Ferncliffe (b) Ingele and (c) Dargle

-+-- buds - opening -Ar- full bloom -)C- withering/fruit set - dead -45-

Experimental pollinations Using individual capitula as the units of replication, fruit set of self-pollinated capitula was significantly lower than that ofcross-pollinated ones at all three study sites, mean number offruits per capitulum for the self-pollinated capitula was zero (Dargle: p

For the cross-pollinations, the percentage capitula which set fruit and the mean number offruits per capitulum was higher at Femcliffe than at any other site (Fig. 3.6 a & b). Furthermore, only at Femcliffe was the mean number offruits per capitulum greater for cross-pollinations than for the open pollinated control flowers (natural seed set) (Fig. 3.6 a & b). Greater fruit set in the hand pollinated capitula is the expected result since these flower-heads received a greater pollen load. The population ofD. odorata at Ingele Forest is one ofthe largest in Natal. High fruit set for the open pollinated (control) capitula at this site may be the product of increased pollinator activity and/or greater within patch genotypic diversity (and therefore more compatible genotypes). Low fruit set for the hand-pollinated capitula at this site could be the result of unsuccessful pollen transfers. Ingele forest is situated at Harding, over 200 km away from the University ofNatal, and trips to this site could only be made every 10-14 days. It is possible that fewer visits and therefore hand pollinations at this site meant that artificial pollen transfers did not always take place when stigmas were receptive.

Pollinators

Delairea odorata is pollinated by generalist insects and the pollination system is unspecialised; the insects transfer pollen from many parts of their body as they forage throughout the inflorescences ofa patch. The most frequent visitors to the inflorescences were Syrphidae (hover flies) and Calliphoridae (Stomorhina sp.) (Table 3.2). Although Apidae (bees) were less frequent visitors to the capitula than both the Syrphidae and Calliphoridae, they carried by far the most D. odorata pollen on their bodies (Table 3.2). A few wasps and beetles were collected foraging on the flowers, however, no pollen was collected from their bodies and they were generally -46-

a

1.4 . _ , -~ • ..~ .._._.. -- _.

1.2 E :J ....:J '0.. t1l ....o ID 0.8 Q. (/) ~ :J.... 0.6 '+- 0 c c 0.4 t1l ID E 0.2

0 Dargle Ingele Ferncliffe study site

~ Cross-pollinated El Natural seed set b 60

852 50 ,-...... ,'#- ~ 40 :J '+-...... c.... .~ 30 t1l .a '0.. 20 ot1l 10

0 Dargle Ingele Ferncliffe study site

~ Cross-pollinated • Self-pollinated El Natural seed set

Fig. 3.6 a & b. Results of experimental pollinations on D. odorata capitula expressed as (a) the mean number of fruits per capitulum (error bars represent 95 % confidence intervals) and (b) the percentage capitula which set fruit (numbers refer to the total number of capitula used in the test). -47­ Table 3.2. Details of the pollen collected from the major pollinators of D. odorata: the number of species of each major group (family), the number of individuals, within each family, with D. odorata pollen, and the average number of pollen grains per insect, is indicated (n=number of individuals collected).

Pollinator (n) number De/airea De/airea Av. number of Pollinator species pollen + other pollen grains frequency only species (±SD) per insect

Apidae (n=12) 1 11 1 622 (250) occasional

Syrphidae (n=33) 9 13 20 115(100) frequent

Muscidae (n=4) 4 1 3 20 (15) occasional

Sarcophagidae (n=10) 1 0 1 15 (13) occasional

Tachinidae (n=3) 3 2 1 25 (9) rare

Calliphoridae(n=31 ) 1 4 27 24 (20) frequent

.HP E ...... - -- - OEC :::l 3 :J ..-0.. ~CC ~ 2.5 .... Q) 0­ 2 .--.-..-----.....-- ..- .-...--...-...------.---.-- -.--...------CIl .;t:: :::l ...... 1.5 - - - - -~ . ------_ . -.._..- --..------.. ._------.-- ci r::: r::: 1 m Q) E 0.5 o Dargle Femcliffe

study site

Fig. 3.7. Mean number of fruits set per capitulum for pollen limitation experiments carried out on D. odorata capitula. Error bar represents 95 % confidence interval. HP: hand-pollinated capitula; EC: open pollinated capitula on experimental plants; CC: open pollinated capitula on control plants. -48- infrequent visitors to D. odorata. Several species ofbutterfly (Hyalites esebria esebria Hewitson, cassius Godart, Precis octavia sesamus Cramer, Vanessa cynthia cardui L., Anthene definita definita Butler and Belenois zochalia zochalia Boisduval) were also observed to move from flower to flower within a patch . However, no pollen was found on their bodies and therefore their role as pollinators ofD. odorata is questionable. The importance of bees and hoverflies in the pollination of the closely related genus Senecio has been reported elsewhere (Gross and Wemer, 1983; Lawrence, 1985).

Pollen limitation The results for the three pollination treatments, at each site, were significantly different (Femcliffe: H=237.75, p<0.05; Dargle: H=63.48, p<0.05). At both study sites, the average number offruits set per capitulum was higher for those receiving extra pollen than for control flowers on the same ramets (Fig. 3.7). The hand-pollinated flowers (HP) therefore received more pollen and also more ofthe plant's resources than did their respective controls. Forthe control flowers on experimental plants (EC) the fruit set per flower was similar to that ofthe control flowers on control plants (CC). The hand-pollinations therefore increased the amount ofresources allocated to seeds at no expense to the other flowers. At Ferncliffe, relative to the controls, hand-pollination increased fruit set by 139 % (BC) and 173 % (CC) whilst fruit set between the two controls differed by only 14%. At Dargle fruit set for hand-pollinated flowers was 81 % greater than that ofthe experimental control and 145 % greater than the control flowers on control plants. The results ofthe open crosses can be used as measures ofthe maximum potential seed set ofD. odorata at each site.

3.4 Discussion

The results ofthe experimental pollinations indicate that D. odorata is an obligate outbreeder. A self-compatibility index (SI) was calculated for each population by dividing the proportion of seeds set from self-pollinations by those in cross-pollinations. A species is self-incompatible ifthe index is less than 0.2 (Bawa, 1974; Zapata and Arroyo, 1978; Sobrevila, 1989). Using this measure ofincompatibility, all three populations are self incompatible since their index was less than 0.2 (Dargle:0.016; Ingele:0.004; Ferncliffe :O). Furthermore, per capitulum, D. odorata has a high pollen ovule ratio of about 3000. High pollen ovule ratios and showy capitula are commonly regarded as indicators ofoutbreeding (Lawrence,1985). -49-

Self-incompatibility, the inability of a plant with functional gametes to produce selfed seeds, occurs widely amongst flowering plants and is known from a least 71 families. It is one ofthe major outbreeding mechanisms in plants, having been recorded from 250 ofthe c. 600 genera studied. Incompatibility results from the inhibition of pollen tube growth (Brewbaker and Majumder, 1961; Richards, 1986). This inhibition commonly occurs at one oftwo sites: (1) on the stigma or soon after pollen tube germination and (2) in the pistil during the first few hours ofpollen tube growth. Based on the site ofreaction two major types ofincompatibility systems are recognised in plants (Richards, 1986): plants having the sporophytic system normally show inhibition on the stigma, while plants ofthe gametophytic-type have pollen tube inhibition in the style or ovary.

Members of the Asteraceae mostly have sporophytic incompatibility systems (Crowe, 1954; Brewbaker and Majumder, 1961; De Nettancourt, 1977;Lawrence, 1985; Richards, 1986; Abbot and Forbes, 1993). Several species of Senecio are self-incompatible (for example: Senecio squalidus L., S. laurus Urv., S. gregorii F. Muell and S. odoratus Hornem) and are likelyto have asporophytic incompatibility system (Lawrence, 1985; Abbott and Forbes, 1993). Homomorphic sporophytic systems are characterised by a single multi-allelic locus and prezygotic mechanisms (De Nettancourt, 1977). In contrast, most ofthe complex genetic systems such as multifactorial incompatibilities and post-zygotic mechanisms are associated with gametophytic systems (De Nettancourt, 1977). In sporophytic systems crosses between individuals will result in 0-100 % seed set, depending on whether they share the same incompatibility alleles or not. Such a contrasting pattern expected in species with sporophytic systems is more easily detected than the complex pattern expected in gametophytic systems (Sobrevila, 1989). In addition, because the expected pattern is so well defined, it might be detectable even under field conditions, where environmental factors that affect seed set cannot be easily detected (Sobrevila, 1989). For this study, it was noted that many ofthe capitula which were hand-pollinated set no seed at all, while for others seed set was frequently above 70 %.

Although the results ofthis study indicate that D. odorata possesses a strong sporophytic self­ incompatibility system, it cannot be ruled out that under certain field conditions the effectiveness of this system will be weakened resulting in an increase in the level of self-compatibility. Breakdown ofsporophytic incompatibility mechanisms can occur (Richards, 1986). Research has -50-

shown that delayed pollination, high temperatures, high CO2 concentrations and electrical stimuli can break incompatibility mechanisms (Richards, 1986). Roggen and van Dijk (1972) found that stigmatic abrasion with a wire brush also affected compatibility. They suggested that damage to the cuticle of the stigmatic papillae promoted penetration of selfed pollen tubes. This may account for the few seeds set in the selfpollination trials ofthis study. Alternatively, as it is not known whether the pollination bags used completely excluded pollen, the few fruits produced by D. odorata from self pollination treatments may be due to inadvertent cross-pollinations or interference by small crawling insects e.g. thrips.

Delairea odorata does not conform to Baker's (1965) "ideal weed" model because it is self­ incompatible. However, it is important to note that Baker (1965) based most ofhis analyses on agricultural weeds and while many ofhis "ideal weed" characters hold true for these plants they do always hold for invasive plants ofnatural habitats (Rejmanek, 1989; Cronk and Fuller, 1995; Mack, 1996). For example, ofthe environmental weeds listed by Cronk and Fuller (1995), 13% are dioecious and 11% monoecious (and the rest hermaphrodite). A similar level of dioecy is found in the native flora ofNewZealand (12-13%) which is a flora considered to have a high level ofdioecy (Cronk and Fuller, 1995). A rather high percentage ofinvasive plants thus appear to be dioecious or monoecious - two mechanisms which promote cross-pollination and therefore outbreeding. Williamson and Fitter (1996) compared attributes ofthe native and invasive species ofthe British flora to determine if any sets ofcharacters were common to invading species. They found that most characters, particularly the "ideal weed" characters, failed to relate to invasion success.

For the pollen-limitation trials, the flowers that were given additional cross-pollen were able to produce nearly twice as many achenes per capitulum than those which were open pollinated (both EC and CC). These results indicate that seed set in D. odorata may be pollen limited. As previously stated, the occurrence ofpollen limitation may occur in a number ofways: (1) low numbers ofpollinators (2) profuse flowering so that the pollinators become satiated (3) ineffective pollen transfer, and (4) restricted gene-flow which may have severe effects on seed fitness, particularly in self-incompatible clonal species, where a single clone may occupy a large patch (Goldingay and Whelan, 1990; Widen and Widen, 1990). -51-

Pollen limitation ofseed set in natural populations has been documented in several studies (for a review see Rathcke, 1983). Widen andWiden (1990) found that distance-dependent fecundity could have severe effects on seed fitness in clonal species occurring in large patches. In the present situation pollen limitation was most likely due to low frequency ofvisits by pollinators and/or high density oframets ofa single clone occupying a patch. However, seed-set for the open pollinated controls, particularly at Femcliffe, does show that patches are likely to comprise more than one genet since, despite the expected movements of pollinators between very near neighbours, some seed was set. Thus low seed set may indicate a general lack ofpollinators. An isozyme analysis or genetic 'fingerprinting' would need to be carried out to confirm the genetic diversity ofpatches. -52-

Chapter 4­ Seed biology

4.1 Introduction Seed biology ofinvasive species Seeds are the product ofsexual reproduction, and are fertilised mature ovules comprising embryos and nutritive tissue, either as endosperm or food stored in the cotyledons (Abercrombie et aI., 1990). Survival ofmost species, particularly annuals where reproduction is only by seed, depends on the production ofa sufficient number ofsexually produced propagules (Zimdahl, 1993). The progeny ofspecies that reproduce from sexually produced seed may be different from their parent plants and therefore these species have the potential to extend their geographical range of distribution. Many invasive species reach reproductive maturity relatively early and produce copious amounts of seed (Zimdahl, 1993; Cronk and Fuller, 1995; Holm et al., 1997). Baker (1965) associated rapid sexual maturation, highfecundity, seed vagility, unspecialised germination requirements or dormancy and seed longevity with weedy potential. Other studies have supported the "ideal weed" characters listed by Baker (1965) (Arnor and Piggin, 1977; Klingman et al., 1982; Stephens, 1982; Deanetal., 1986; Zimdahl, 1993) and, although most ofthese studies were based on weeds of agricultural environments, similar reproductive traits have been found to be common to the invasive species of natural habitats. Amor and Piggin (1977) studied factors influencing the establishment and success of exotic plants in Australia and concluded that characteristics common to invasive species included rapid seed production after anthesis, high seed production under a wide range of conditions and dormancy mechanisms preventing germination under unfavourable conditions.

The Asteraceae is one ofthe most important weedy families worldwide (Heywood et aI., 1977; Heywood, 1989; Cronk and Fuller, 1995). The success ofthis group has been partly attributed to a high reproductive rate, the development of which is largely associated with the evolution of the capitulum (Heywood et aI., 1977). The aggregation ofreduced flowers into a single head is sometimes associated with geitonogamy (often superimposed with agamospermy), pollinator attraction and the production of single-seeded fruits (achenes) with effective dispersal mechanisms. The one-seeded fruit often has a pappus (ring offine hairs) or wings to aid in wind -53- dispersal. Hooks and spines frequently aid dispersal by animals and man (Heywood et al., 1977; Heywood, 1993). Most of the weedy species of this family mature early and produce large numbers ofsingle-seeded, dry fruits often with unspecialised germination requirements (popay and Roberts, 1970; Lawrence, 1985; Holm et a!., 1997). Among the weedy members of the Asteraceae listed by Holm et al. (1997) many (62 %) are annuals, reproducing principally by wind dispersed achenes, and some are perennials relying heavily on sexual as opposed to vegetative reproduction. For example, Senecio vulgaris L. (an annual) and Chondrilla juncea L. (a perennial), both serious economic weeds, are described as "prolific seeders" with rapid seedling establishment (Holm et a!., 1997). Despite a mild dormancy in a small.portion ofthe seeds of these species, most are capable of immediate germination. Seeds of C. juncea and S. vulgaris germinate in both the light and dark and the optimum temperature for germination is 25 0 C (popay and Roberts, 1970; Holm et a!., 1997). Lawrence (1985) analysed characteristics of32 Senecio species in Australia and found that 29 species reproduced sexually, producing over 10 000 small achenes (0.1-0.6 mg) per plant. Few species showed specific germination requirements.

General biology ofweed seeds and seedlings The germination process involves the inception of rapid metabolic activity within the seed, resulting in perceptible growth ofthe embryo. It is usually associated with the uptake ofwater and oxygen, use ofstored food and normally, release ofcarbon dioxide (Klingman et a!., 1982; Zimdahl, 1993). For many investigators radicle emergence has been one ofthe best visible criteria identifying the initial stage ofgermination (King, 1966 ; Evetts and Burnside, 1972; Robocker, 1977; Grime et a!., 1981; Smreciu et al., 1988). For a seed to germinate it must have an environment favourable to the germination process, factors promoting germination may include specific temperature, light, moisture and pH requirements (King, 1966; Klingman et al., 1982). Generally, weeds have no special requirements for germination (Baker, 1965; Stephens, 1982; Zimdahl, 1993). However, in a few species dormancy mechanisms do occur (Zimdahl, 1993). Five environmental factors affect seed dormancy: temperature, light, moisture, oxygen and the presence ofinhibitors (including allelopathic effects). Other factors directly related to the seed and its dormancy include impermeable seed/fruit coats, immature embryos and an after-ripening period (King, 1966; Grime et a!., 1981; Klingman et al., 1982; Zimdahl, 1993; Cousens and Mortimer, 1995). -54-

While there are great differences in the rate ofseedling growth among the various weed species, the most competitive usually have rapid growth and early, extensive root development compared to the plants with which they are competing (King, 1966; Zimdahl, 1993; Cousens and Mortimer, 1995, Holm et a!., 1997). They grow tall quickly or gain competitive advantage by twining up larger plants. Furthermore, they are frequently tolerant of shade in that their highest carbon dioxide assimilation is not in full sunlight (Zimdahl, 1993). Certain weed seedlings have large expansive type foliar cotyledons that, through early photosynthetic function enable young seedlings to become established quickly. This is true for the Asteraceae (King, 1966). Similarly, the rate ofnew leafproduction is important, as it affects surface area available for photosynthetic production.

The present study In southern Africa, Hawaii and AustraliaD. odorata reproduces sexually by seed and vegetatively by stolons. However, in California, where this vine is a serious weed, no viable seed is produced and the plant reproduces only by vegetative means (Balciunas, pers. comm.). In a study conducted by the USDA Agricultural Research Service on the germination requirements ofthe seeds produced by D. odorata in California, for the entire testing procedure, only one seed germinated (USDA, 1998). The reason for the sterility of the populations in California is at present unknown (USDA, 1998). The contrasting modes ofreproduction ofD. odorata in exotic infestations raises interesting questions as to the nature of the spread of this vine. Sexual reproduction is usually an important method ofpropagation for weeds ofthe Asteraceae; they generally mature early and produce copious amounts ofeasily dispersed single-seeded fruits which germinate quickly or remain buried beneath the soil until germination requirements are met. Frequently, it is because ofthe large amount of seed produced that eradication and control of these weeds is difficult. However, if D. odorata is a serious invader in California despite its lack ofviable seed then the importance ofthe seed biology of this vine in the process ofinvasion needs investigation.

Besides the investigation conducted by the USDA Agricultural Service (California) into seed germination, no studies have been conducted into the seed biology ofthe fruits ofD. odorata. Without an understanding of the seed biology ofD. odorata in southern Africa, Hawaii and Australia no predictions about the importance ofsexual reproduction in the spread ofthis species -55-

in·exotic locations can be made. Furthermore, if this species does spread through sexually reproduced seed in Hawaii and Australia, knowledge ofthe seed biology may be important in the development and implementation ofmanagement strategies in those areas.

The present investigation was conducted to obtain a better understanding ofthe seed biology of D. odorata in southern Africa. It is a baseline study, the results of which may be used as a comparative reference for studies made on the seed biology ofthis species in exotic locations. A series oflaboratory experiments were conducted to determine the germination characteristics of the seeds. An examination of the effect of light on seedling development was also made . Although the difficulties in relating laboratory experiments to field conditions have been pointed out (Koller and Roth, 1964; Popay and Roberts, 1970), such analyses are an essential beginning to an understanding ofthe behaviour ofseeds/fruits in response to the complex and fluctuating conditions in the field (popay and Roberts, 1970).

4.2 Materials and methods All germination experiments were conducted on mature achenes of D. odorata. Mature infructescences, collected in June 1998 from Ferncliffe Nature Reserve, Pietermaritzburg, were brought back to the laboratory where the mature fruits were removed . The achenes were dry

stored in the dark at room temperature (20 0 C) for one month prior to germination trials.

Germination requirements

Allgermination studies were conducted inthe laboratory. For alltreatments, achenes were placed on moist filter paper in petri dishes. Germination was recorded every second day and was considered to have occurred once the radicle had elongated to a length of2 mm or more. Unless otherwise stated, photoperiod was set at 16 h light/8 h dark and 10 replicates of20 achenes each were used per treatment.

1. Fruitweight andmeasurements:As fruits are very small, average weight was determined by massing the achenes in batches of20 (20 replicates). The average length and width ofthe fruits was determined using a dissecting microscope (50 achenes). 2. Pericarp permeability: The ability ofD. odorata achenes to imbibe water was established

by incubating the mature fruits at 25 0 C on moist filter paper in petri dishes. Fruits were -56-

blotted dry before massing at 24-hourly intervals for 12 days. 3. Temperature requirements: The influence oftemperature upongermination rate was studied at 5 constant temperatures: 10°C, 15 -c, 20°C, 25 -c and 30°C. 4. Responseto light: Light requirements were determined by comparinggermination ofachenes in petri dishes wrapped in aluminum foil with those exposed to light. Achenes were germinated in the dark at 10°C, 15 "C, 20°C, 25 "C and 30°C. Radicle protrusion from these fruits was examined in a growth room fitted with a green "safe light". 5. Response to chilling: Following one month storage in the dark at 0 °C, the effect ofchilling on viability ofachenes was determined by germinating achenes at 15 °C. Fruits were kept on moist filter paper in petri dishes covered with foil during chilling. 6. Longevity:Viability ofseeds following one year dry storage, in the dark at room temperature (20 °C), was tested by germinating seeds at 10, 15 and 25°C both in the light and dark. 7. Depth ofplanting : To establish the effect ofburial depth on achene germination, achenes ofD. odoratawere placed in plastic pots and buried at depths of0 cm, 1 cm and 5 cm below the soil surface. The pots were kept in the greenhouse maintained at 20-25 "C and the soil was kept moist. Seedling emergence was recorded after seven weeks; three replicates of40 achenes each were used per treatment.

Seedling establishment The effect ofshading on the development of seedlings was assessed. Three light treatments, 71 %, 45 % and 17 % ofPAR (photosynthetically active radiation) were employed (see Chapter 5 for full description of materials) . Six replicates of 10 mature achenes were used for each treatment; achenes were placed in pots containing a 50:50 mixture ofsoil and composted bark and the pots were assigned to one ofthe three light treatments. The development ofseedlings was monitored continuously; the following life history characters were recorded for each treatment:

1. days from sowing to first germination (assessed as emergence from the soil); 2. number ofseeds germinating and seedling survival for the first four months following sowing and survival eight months after establishment; 3. plant height (mm) and leaf number for the first four months following sowing and eight months after germination;

4. length (mm) ofthe longest five leaves for each seedling eight months after germination; and -57-

5. root, shoot and leaf dry weight (g) eight months after germination. On harvesting each seedling was separated into roots, leaves and stems and placed into paper bags. The plant components were air-dried with heater fans (± 40 QC) for approximately four days and the dry weight of roots, stems and leaves was recorded.

Data analysis Allpercentage values were arcsinetransformed prior to analysis (Zar, 1974). This transformation is applicable to binomial data expressed as decimal fractions or percentages, and is especially recommended when percentage values cover a wide range ofvalues (Steel and Tome, 1980). For all normal data one way ANOVA was used to test for significant differences between treatments. Duncan's Multiple Range Test was used to compare means. Non-normal data was analysed using the non-parametric ANOVA equivalent, the Kruskal Wallis Test (Zar, 1974). The statistical programme used was Minitab? (1995).

4.3 Results Germination requirements The size ofthe achenes ofD. odoratawas assessed by weight (mg), length (mm) and width (mm). As is the case in many members of the Asteraceae (Lawrence, 1985), the achenes ofD. odorata are very small (Table 4.1). The water uptake study, conducted to determine the permeability of the pericarp, indicated that water was imbibed almost immediately (Fig. 4.1). Over 12 days the weight ofthe achenes increased from 5.1 ± 0.5 mg to 39.8 ±13.4 mg.. All ofthe achenes which imbibed water germinated indicating that the fruits were non-dormant.

The germination trials conducted in the light and dark at 10, 15,20,25 and 30 QC indicated that germination occurred, in both the light and dark, over the temperature interval of10-25 QC (Figs 4.2 & 4.3). No germination occurred, in the light or dark, at 30 QC (Figs 4.2 & 4.3). At the temperatures 10, 15, 20 and 25 QC there was no significant difference in final germination percentage ofseeds exposed to light (F=1.4, p>0.05). In the dark, final percentage germination at 25 QC was significantly different from that at lower temperatures (F=4.25 , p<0.05, Duncans Multiple Range Test). In the light, maximum germination occurred at 25 QC; an optimum temperature for the germination of many species of the Asteraceae (Lauer, 1953; Popay and Roberts, 1970; Forsyth and Brown, 1982; Holm et al., 1997). Although germination was not inhibited in the dark, for these trials germination values were generally lower and the optimum -58~

Table 4.1. Average weight and size of achenes (±s.d.) of D. odorata.

Measured Measurement parameter

weight (mg) 0.3 (±0.03)

length (mm) 1.5 (±0.16)

width (mm) 0.6 (±0.09)

60

50 ...... , Cl ...... E 40 IDen C1l ID '- 30 / 0 c en / en /' C1l 20 E ~ 10 ..L ~ T ...,.. 0 o 1 2 3 4 5 6 7 8 12

incubation time (days)

Fig. 4.1. Increase in mass (water uptake) of fruits of D. odorata imbibed at 25 °C for 12 days. Photoperiod was set at 16 h light! 8 h dark. Error bars represent standard error. -59- 100

80 I=~~

~ 0 -c: 60 0 :;:: III c: .§ 40 Q) Cl 20

0

incubation time (days)

Fig. 4.2. Cumulative % germination of D. odorata achenes maintained for 15 daysat one of five temperatures. Photoperiod was set at 16 h light! 8 h dark. Error bars represent standard error. LSD = 12.5

- x- 10 DC __ 15 DC -A-20 DC -.-25 DC 30 DC

100 ., .

80I=~? . . -.... .t···I··-I ···· III .x-x-x I____x-x-)(' c 60 ..-..-.-- __.--- o :;:: III .5 ....E 40 -----.- ..-- Q) Cl

20 -j ...... • f ··· ···· ························· ·······X · ·· ······ ···.....•. •......

1 3 5 7 9 11 13 15 17 19

incubation time (days)

Fig. 4.3. Cumulative % germination of D. odorata achenes maintained for 20 days in the dark at one of five temperatures. Error bars represent standard error. LSD=12.05

- x- 10 DC __ 15 DC -A-20 DC -.-25 DC 30 DC -60- temperature for germination was also decreased, maximum percentage germination was at 10 "C. Although the seeds placed at 30 "C did not germinate, when these seeds were placed at 15 "C

(after 12 days exposure to 30 o C) germination occurred indicating that the seeds were still viable. Chilling for one month in the dark at 0 "C caused a slight increase in time to germination, in the light and dark, at 15 "C, but had little effect on percentage germination at this temperature (Fig. 4.4). Fruits ofD. odorata, set in June 1998, were dry stored for one year in the dark at room temperature (20 CC) and subsequently sown, in the light and dark, at one ofthree temperatures­ 10, 15 or 25 cc. Results show a considerable decrease in the viability of these seeds at all temperatures (Fig. 4.5). No seeds germinated at 10 "C and, for all treatments, maximum germination did not exceed 40 % (cf Figs 4.2 & 4.3) .

The results of experiments conducted to determine the effect ofburial on achene germination showed that only achenes scattered on the soil surface were capable ofgermination (Table 4.2) . This finding is in accordance with the findings ofBeveridge and Wilise (1959) who recorded that larger seeds emerge more easily from greater depths in the soil than smaller seeds. With an achene mass ofonly 0.26 mg the result obtained for the achenes ofD. odorata was thus expected.

Seedling establishment In order to examine the development ofD. odorata seedlings in response to light, trials at 71, 45 and 17 % sunlight were conducted. Barring the achenes exposed to 71 % sunlight, seedling emergence was completed within the first three weeks of planting (Fig. 4.6). Maximum percentage germination was significantly greater at 17 and 45 % sunlight than that at 71 % sunlight where only 60 % of the achenes germinated (F=5 .15, p<0.05). For all treatments, survival decreased slightly with time from sowing. There was no significant difference in percentage survival following eight months exposureto the different light levels (F=O .17, p>O .05).

For the first three months, the general increase in seedling growth was arrested (Figs 4.7 & 4.8). This may be due to an experimental effect, alternatively lack of shoot development may be associated with extensive root growth, a factor frequently associated with the development of competitive weed seedlings (King, 1966; MeVean, 1966; Panetta, 1977; Zimdahl, 1993; Cousens and Mortimer, 1995; Holm et al., 1997). After four months growth, and until termination ofthe -61-

100 - - ~ . __-- .-_.-.-.------.--.--- .------.--.,-.--. ------.- -..--- -_._-.-._.._-.------.-~ . _------.- _.-.__.------.--.------.- - . ------~ -- - -. _ ------. - - -.------..--..--- --

80 .-..------.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

incubation time (days)

Fig. 4.4. Cumulative % germination of D. odorata achenes following treatment for one month in the dark at 0 °C. Achenes were germinated at 15°C in both the light and dark. Error bars represent standard error. --light _ dark -62- 100

80 ...-. ~Iight ~ 0 -c 60 • dark 0 :;:: ro .!;;; E... Q) 40 Cl

20

0

incubation temperature CC)

Fig. 4.5. Final % germination, in the light and dark, of dry stored (1 yr) D. odorata achenes, following 20 days treatment at 10, 15 and 25°C. Error bars represent standard error.

Table 4.2. Effect of depth of planting on seedling emergence of achenes of D. odorata .

Planting Seedling emergence depth (cm) (%)

o 76 (± 12) 1 0.3 (±0.5) 5 o -63-

120 ~ .

100 ------

80 -1 ··························· ·········· -1+ .-..- T~ · · · · ·············· ··- ·······""", ····-···f ········· ·········· - -.- ...... -..- -- .

=--+---~- -- ·· - · - - -- f - - - - · · -.-- .... .------.--

40 + · ··· ······· ······1 1

O-+- --r---.---~--.----.---.----,---.----, Oct Nov Dec Jan Feb Mar April May Jun

month

Fig. 4.6. Emergence and subsequent survival (%) of D. odorata achenes following eight months treatment at 71 , 45 and 17% sunlight. Error bars represent standard error.

-71% ---45% -.-17% -64-

600 .~ -.---..-.. ~..-.--.- ~ ~.~-~ --.. - --.~ ..--..-- ~-.- ~ . -.-.. ~- .-- _. --- ~ _.~..- -~ .- ~-- . - -..-~-. -~ _..-.~-_.~

E .s .c 400 -_------.--.-.--.-- ~ .- - --'- -.-.-.--..------.--'-'-- - .- -- -- . - - - ~ - - . - . - . - - ~ _ . _ . -- . - -.- -_.._._._.__ _.-_.._._-_._.._.._ _.. - C, c: ~ E ID 1ii 200 -I ,/ . 1 Nov Dec Jan Feb March April May Jun

month

Fig. 4.7. Monthly increase in stem length (mm) of D. odorata seedlings exposed to 71 , 45 and 17 % sunlight. Error bars represent standard error. - 71% --45% ---17%

40

30 -1 . . . .

'0 20 · ·------.----..--.- ---..--- ______-._.__..---.------.-- -.------.------_. _-.'. .... ID .0 E ~ c 10 --- -

0 +---....---....----.-----='=--....----.--_ _ -.--_ _ -.--_--. Nov Dec Jan Feb March April May Jun

month

Fig. 4.8. Monthly increase in leaf number of D. odorata seedlings exposed to 71, 45 and 17 % sunlight. Error bars represent standard error. -71% - - 45% ---17% -65-

experiment, stem length ofseedlings exposed to 17 % sunlight was significantly greaterthan those grown at 45 and 71 % (Kruskall Wallis ANOVA, p< 0.05) (Fig. 4.8) . Number of leaves per seedling was maintained at a relatively constant level in all treatments for the first 12-15 weeks (Fig. 4.8). However, after eight months the average number of leaves per seedling was significantly greater for those seedlings grown at 45 % sunlight (H=8.51, p<0.05) (Fig. 4.8). There was no significant difference in leaf size of seedlings exposed to the different light treatments, although leaves ofseedlings exposed to only 17 % sunlight were generally larger than those grown at 71 % (H=1.29, p>0.05) (Fig.4.9).

The flowering period for D. odorata occurs over the months ofApril to June . In June 1999 the seedlings for this investigation were eight months old and they had no flowers . This is contrary to expectation as weedy Sene cio species usually reach sexual maturity rapidly and flower within their first year of establishment (Lawrence, 1985; Abbot, 1986 ; Holm et al., 1997). Further studies are required to confirm the result obtained for this investigation. Eight months after sowing the seedlings were harvested. Present results revealed no significant difference between treatments in the allocation ofbiomass to leaves, stems and roots for seedlings exposed to the different light levels (leaves; F=I .68, p>0.05; stems, F=0.5, p>0.05; roots, F=I .20, p>0.05) (Fig. 4.10) . Maximum biomass allocation was to stems, followed by that to roots and leaves.

4.4 Discussion Germination requirements

The results of this study corroborate those of Grime et at. (1981). Their analysis of the germination characteristics ofseveral plants drawn from a local flora in northern England, found that the majority of species with small seeds germinated easily and were non-dormant. Their investigation also showed that fruits of the Asteraceae have amongst the highest capacity for immediate germination. High initial germinability ofseeds ofthe Asteraceae was correlated with the presence ofthe pappus and antrorse hairs or teeth . Studies have shown that these structures are characteristic of species in which germination normally occurs in seeds lodged on the soil surface (Sheldon, 1974; Grime et al., 1981). Radicle emergence and early seedling establishment ofthese species appears to benefit fromthe combined effects ofseed shape and antrorse hairs or teeth, both ofwhich tend to anchor the fruit in an upright position in loose soil bringing the radicle end into closer contact with the soil (Grime et al. , 1981). -66-

40

30 . _. ... •...... E ...... E .c Cl 20 c ~ ro -~ 10

71% 45% 17%

sunlight (%)

Fig. 4.9. Length of the longest leaves (mm) of D. odorata seedlings following eight months exposure to one of three light levels. Error bars represent standard error .

80 1J71% ~ ~ .45% c 0 17% .Q 60 1§ ,g . m :zro 40 E 0 :0 20

leaves stems roots

plant part

Fig. 4.10. Dry weight allocation (%) to roots, stems and leaves of D. odorata seed lings following eight months exposure to one of three light levels. Error bars represent standard error. -67-

Clearly light stimulates the germination ofD. odorata seeds. Light is one ofthe major factors regulating seed germination and light requirements are frequently associated with small seeds (Harper et al., 1970; Mayer and Poljakoff-Mayber 1989; Bell, 1993; Plummer and Bell, 1995; Rokich and Bell, 1995) . Plummer and Bell (1995) documented a light requirement for germination in several small seeded Australian Asteraceae. Several other authors have also reported a light stimulation within the Asteraceae (Popay and Roberts, 1970; Mott, 1972; Atwater, 1980; Grime et al., 1981; Willis and Groves, 1991). Lack ofgermination ofburied D. odorata seeds may also indicate a light requirement. Light-promoted seed germination appears to be an adaptation to enhance the chance ofestablishment where germination ofdeeply buried seeds would prove fatal. Physiologically active light flux densities rarely penetrate more than a few millimeters into the soil (plummer and Bell, 1995). Small seeds contain limited reserves and it is advantageous for germination to occur where photosynthesis can quickly take over from stored carbohydrates (Pons, 1992; Plummer and Bell, 1995). Plummer and Bell (1995) found that Asteraceae seeds weighing less than 0.5 mg required light, whereas the response was mixed among species with heavier seed.

Lowtemperatures clearly have a beneficial effect on the germination ofseeds ofD. odorata when not exposed to light. Treatment with alternating temperatures or low temperature incubation has long been known to increase germination in darkness oflight requiring seeds (Popay and Roberts, 1970; Bewley and Black, 1994). Grand Rapids lettuce, for example, is generally dormant in darkness above 23 QC, below this value seeds germinate without illumination (Bewley and Black, 1994). The achenes ofD. odorata exposed to 30 QC germinated easily when moved to a lower temperature. This phenomenon is known as thermoinhibition (Horowitz and Taylorson, 1982). Thermoinhibition refers to seeds where germination fails at high temperature but proceeds upon subsequent transfer to an optimal temperature, thermoinhibition thus prevents germination when conditions are unfavourable.

It may be unwise to assume that the ability offreshly collected seed to germinate in the laboratory is a reliable indication that under field conditions germination occurs soon after fruit release. In the field, the germination of freshly dispersed seed may be prevented by limiting factors not operating in the laboratory tests (Grime et al., 1981). Available literature suggests that the germination of achenes of many members of the Asteraceae coincides with the onset of wet conditions (Grime et aI., 1981; Holm et aI., 1997). In southern Africa the small wind-dispersed -68- fruits of D. odorata are set and dispersed in the dry winter months of June and July when average daily temperatures range between 5 and 20 QC (CCWR, 1999). Although the achenes ofD. odorata have no stratification requirement, the cold dry winter conditions may prevent germination in the field. Germination ofachenes may coincide with the onset ofthe spring rains and warmer temperatures oflate September. Seedling germination, establishment and growth in the field will need to be examined to confirm these preliminary observations.

Seedling establishment Competitive capacity ofplants is governed, in part, by the efficiency with which they intercept light (Keeley and Thullen, 1978). In relation to light requirements for plant establishment and growth Grime (1966) made a distinction between shade avoiding and shade tolerant species. The observations made in this investigation indicate that the seeds ofD. odorata germinate at both high and low light intensities, although establishment in lower light conditions (17 and 45 %) appears to be favoured. D. odorata is therefore likely to be a shade tolerant species. Maximum leafproduction occurred at 45 % sunlight, a factor confirmed by other observations made for this project (Chapter 5). In its native environment D. odorata frequents forests and forest margins where variation in light intensity is typical. Furthermore, in areas where D. odorata creates a mat­ like groundcover, extensive die back of the adult plants over winter may expose seeds to an environment ofvaried and changing light intensity throughout the day; an ability to tolerate a range oflight intensities may thus confer a competitive advantage to the seedlings ofD. odorata.

4.5 Concluding comments Rapid growth through the vegetative phase to flowering, prolific seed production, unspecialised germination requirements, seed longevity and dormancy and adaptations for long and short distance dispersal have often been associated with weedy potential (Baker, 1965; Amor and Piggin, 1977; Klingman et al., 1982; Stephens, 1982; Dean et al., 1986; Zimdahl, 1993; Cronk and Fuller, 1995; Holm et al., 1997). Furthermore, characters of late successional species (high competitive ability and shade tolerence) have been associated with species which are successful invaders offorested habitats (Cronk and Fuller, 1995). Results ofthis investigation show that D. odorata is a small seeded species with relatively unspecialised germination requirements, apparently no dormancy mechanisms and only moderate seedling growth potential (no flowers are produced during the first year ofgrowth). However, to confirm the findings ofthis investigation, further studies into the seed biology ofthis species must be conducted. These should include an -69- extensive investigation into the effect of light on seed germination to determine the importance ofthis environmental factor in seedling establishment.

Reproduction by seed, is frequently associated with the colonisation ofnew habitats (Arnor and Piggin, 1977). Thompson (1990) stated that for clonal dominants of tall herb vegetation, in which the main role ofseed is colonisation ofnew habitats, small seed size and low nuclear DNA content are typical. Small mean seed mass is often associated with large seed numbers, efficient dispersal and high initial germinability (Rejmanek, 1995). Low DNA content has been positively correlated with seed size and associated with the selection for short minimum generation time (the duration ofminimum period from germination until the production ofthe first seed) in time limited environments (Bennet, 1972; Bennet, 1987; Thompson, 1990; Rejmanek, 1996). Interestingly, among 32 species ofSenecio analysed in Australia, three exotic invaders (Delairea odorata (syn. Senecio mikanioides), Senecio pterophorus L. and Senecio vulgaris) have significantly lower nuclear DNA content than 28 native and one non-invasive cultivated species (Lawrence, 1985). -70-

Chapter 5 Growth and reproduction in relation to light availability

5.1 Introduction Although weeds usually reproduce by means ofseeds, they also multiply by means ofvegetative or asexual methods (Leakey, 1981; Klingman et al., 1982; Zimdahl, 1993). Rhizomes, stolons, tubers, roots, bulbs and bulblets are all vegetative means of reproduction. In some important perennial weeds reproduction is primarily vegetative, for example Eichhorniacrassipes (Martius) Solms-Laubach, Heiracium floribundum Wimm and Grab, Cirsium arvense L., Ranunculus repens L. and Salvinia molesta D.S . Mitchell (Leakey, 1981; Eriksson, 1989; Zimdahl; 1993; Cousens and Mortimer, 1995; Cronk and Fuller, 1995). In an analysis of the taxonomic distribution of invasive angiosperms Daehler (1998) found rapid vegetative reproduction, especially in woody plant species, to be a good indicator ofinvasiveness. Vegetative structures can act as a meristem (bud) bank, akin to a seed bank, capable of producing shoots when stimulated to do so. Since a single shoot may produce many subterranean structures and each of these may produce a number of aerial shoots, an extensive clonal stand can arise from a single seed or vegetative propagule (Cousens and Mortimer, 1995). Thus, in contrast to a population ofunitary plants, such as annual weeds, a clonal plant population typically consists ofa limited number ofgenotypes, the 'genets', each represented by a number ofcopies, the 'ramets' (Kays and Harper, 1974; Angevine and Handel, 1986). Hypotheses on the adaptive benefits ofvegetative (clonal) reproduction in plants have two general themes. One emphasises how clonal growth may allow the individual to harvest patchily distributed resources and predicts a relationship between the clonal growth form and competitive dominance (Lovett Doust 1981; De Steven, 1989). The other emphasises clonal proliferation as a means of spreading mortality risk to the individual among many vegetative propagules capable ofindependent survival (Cook, 1979; Cook, 1985; De Steven, 1989) . Because oftheir greater energy reserves, shoots from vegetative organs are often better competitors than seedlings and thus allow for rapid expansion ofa plant in the local environment (without the need for sexual reproduction) (Zimdahl, 1993; Cousens and Mortimer, 1995). Whereas sexual reproduction, usually associated with long distance dispersal, expands the geographic range ofa species, vegetative reproduction contributes to parochial persistence ofthe population (Eriksson, 1992; Cousens and Mortimer, 1995). -71-

It is common for the availability ofresources, such as light and nutrients, to vary in the field and this variation can have a profound effect on both the performance ofindividuals and the dynamics of plant populations (Harper, 1977; Hutchings and De Kroon, 1994; Evans and Cain, 1995). Phenotypic plasticity, the capacity for differential expression of a phenotype owing to environmental influences on the genotype, has long been considered a major means ofindividual adaptation to environmental heterogeneity and has also been associated with invasive and weedy potential (Zimdahl, 1993; Hutchings and De Kroon, 1994; Mack, 1996; Vermeij, 1996). More recently, plasticity has also been viewed as a behavioural phenomenon allowing many vegetative (clonal) plants to arrange their ramets within the environment in a selective manner (Slade and Hutchings, 1987 a-c; Hutchings, 1988; De Kroon and Hutchings, 1995; Evans and Cain, 1995). For example, the ability of a stoloniferous clone to exploit favourable patches or avoid unfavourable patches has been described as "foraging" behaviour (Bell, 1984; Hutchings, 1988; Evans and Cain, 1995). Clonal foraging in a heterogeneous environment is controlled by plasticity in one or more of the following traits: branch production, internode distance (spacer length between ramets) and angle of rhizome growth. Morphological plasticity in these traits is hypothesised to be adaptive in environments where there is high spatial resource heterogeneity (Evans and Cain, 1995) .

Many plants display extensive foraging, in terms ofplasticity in growth form and deployment of biomass, in response to nutrient and light availability (Pitkella et al., 1980; Slade and Hutchings, 1987 a & b; Hutchings, 1988; Sutherland and Stillman, 1988; Salonen, 1994; Hutchings and De Kroon, 1994; Evans and Cain, 1995; De Kroon and Hutchings, 1995) . Light is a limiting resource in many environments and varies both spatially and temporally, affecting plant growth and reproductive output (Anderson et aI., 1969; Pitkella et al., 1980; Slade and Hutchings 1987b; Mitchell and Woodward, 1988). The effects of light on plant morphology, physiology and resource allocation in different light environments have been extensively studied and are well understood (Bjorkman, 1980; Fitter and Hay, 1981; Mooney and Chariello, 1984). In a patchy environment, such as a forest edge or beneath the forest canopy where light is unevenly distributed, the clonal growth form is advantageous as it allows plants to place ramets selectively in bright patches.

Theoretical models predict that under low light conditions plants will show reduced branching, -72- longer internodes and a change in the allocation ofbiomass to different structural components (Slade and Hutchings, 1987b; Solangaarachchi and Harper, 1987; Sultan and Bazzaz, 1993; Salonen, 1994). The predicted adaptive response of leaves to low light is an increase in the relative biomass allocation to these organs and to produce "shade leaves", which differ in area, structure and photosynthetic capacity to "sun leaves" (Sultan and Bazazz, 1993). Light has been cited as a factor affecting allocation of biomass to sexual or vegetative reproduction with sexual reproductive effort increasing with light availability (Anderson et aI., 1969; Thompson and Willson, 1978; Pitkella et al., 1980; Sultan and Bazzaz, 1993).

The present study Delairea odorata expands vegetatively through the spread of stolons and frequently sends out runners in response to poor light conditions (Chipping, 1993). Along a stolon, each node has the potential to root and become a new plant, a ram et, genetically identical to the parent plant (Chipping, 1993; Archbald, 1995) . In addition, thick rhizomes support regrowth when shoots are damaged or removed (Chipping, 1993; Archbald, 1995; Alvarez, 1997). Vegetative propagation ofD. odorata is achieved by ramets which are produced at every node along a principal stolon which creeps over the surrounding vegetation or soil surface. Along the principal stolon, further stolons can develop from the axillary buds present at each node, these will grow to produce a further set ofgenetically identical ramets. For stoloniferous species Hutchings (1988) described the principal stolons as the primary stolons, whereas the axillary buds give rise to the secondary (and tertiary) stolons and ramets. With this kind ofgrowth pattern the plant ultimately develops as a branched, inter-connected population ofindividual clones able to cover a large area in a small amount oftime (Slade and Hutchings, 1987b).

Because ofthe usually large amount oftemporal and spatial variation in light availability, plants living in forests or forest edges can be expected to respond to this environmental factor in terms oftheir growth and reproduction. SinceD. odorata grows primarily in moist, shady environments along forest edges, the factors affecting its growth and reproduction are of interest, especially as very little has been published on this subject. An experiment was designed to determine the response of vegetative propagules of D. odorata to light intensity in terms of growth rate, morphology and resource allocation. -73-

5.2 · Materials and methods All material used in this experiment originated from a single population ofD. odorata located at Ferncliffe Nature Reserve, Pietermaritzburg. To eliminate the effects of genetic variation all cuttings were taken from stolons collected from a single 2 m x 2 m plot. One hundred and eighty single node vegetative cuttings (2-3 cm) were placed into plastic pots containing a 50:50 mixture of composted bark and soil and subsequently transferred to a misthouse for acclimation and establishment. After 10 days all cuttings had rooted and produced new shoots (2.5 ±0.5 cm). These established vegetative cuttings were used in the light experiment.

To determine the effect of light on growth and resource allocation patterns of D. odorata, vegetative cuttings, sixty per treatment, were grown at 71, 45 or 17 % available sunlight (pAR­ photosynthetically active radiation). Metal frames (1.6 m x 1.3 m x 0.6 m) covered with shade cloth were placed over the pots to form shade tents which provided the required light levels. The

2s-1 difference in photosynthetically active radiation (pAR:umol m- ) incident upon transplants for each treatment was recorded with a Skye SKP 200 PAR meter. Mean midday PAR incident on transplants during the experimental period for each treatment varied between 46.9 and 189.5 umol m's' (Table 5.1). These light levels encompassed the range of natural light levels experienced in the field.

2s 1 Table 5.1. Mean PAR (umol m- - ) [±s.d.] incident on transplants in each treatment.

available PAR (±s.d.]

2s-1 sunlight (%) (umol m- )

71 189 [9.1]

45 120 [5.1]

17 46 [3.7]

After 12 weeks exposure to these light regimes the following variables were recorded for each treatment: Vegetative characters (n = 60 for each treatment): leaflength (mm), leafwidth (mm) and petiole length (mm) for 10 mature leaves per plant; number ofleaves; length ofprimary (1 0) stolon (cm), length of secondary (2°) stolon(s) (cm) and length of the first 10 sequentially produced internodes (cm). Resource allocation analyses: 28 transplants from each treatment were harvested, air dried for 72 hours, and weighed. The dry weights ofthe leaves, roots and -74-

stolons (both 1°and r stolons) were determined separately for each plant. Total biomass was calculated for each plant as the summed dry weights of all plant parts. Proportional biomass components were calculated for the roots, stems and leaves. The proportion of stem biomass allocated to secondary stolon production for each transplant was also calculated.

The characters described above are viewed as growth characters indicative ofplant development and fitness due to growth under different environmental conditions and have been used by other researchers (Pitkella et al., 1980; Slade and Hutchings 1987b; Sultan and Bazzaz, 1993; Salonen, 1994) .

To determine the effect of light on allocation of resources to sexual reproduction, plants not harvested were maintained in their respective shade tents for a further four months - until the flowering season in May-June. At peak flowering these transplants were harvested and the biomass allocated to sexual reproduction was determined.

Data analysis

For all normal data one way ANOVA was used to compare differences between treatments. Duncan'sMultiple Range test was used to compare differences between means (Zar, 1974) . For data which were non-normal the non-parametric (ANOVA equivalent) Kruskal Wallis test was used to compare aspects ofgrowth under the different treatments. Data were analysed with the statistical package Minitab",

5.3 Results

There was no significant difference in shoot length, number ofleaves and leaflength ofcuttings at the start ofthis investigation (shoot length: F=1.45, p>0.05; no. leaves :F=0.64, p>0 .05; leaf length: H=2. 14, p>0.05).

Leaves

After 12 weeks, mean leaf length and width differed significantly between plants grown at different light intensities (leaflength,F=58.3, p<0.05; leafwidth,F=45.14, p<0.05). Plants grown at 45 % available sunlight produced the largest leaves, followed by plants grown under heavy shade (17 % PAR). The smallest leaves were produced by those plants exposed to 71 % available -75- sunlight (Fig. 5.1). The same trend was observed for petiole length (H=46A2; p<0.05) (Fig. 5.2a). Light level also had a significant effect on leafproduction per plant, plants exposed to 45 % PAR had more leaves than those grown at both 71 % and 17 % available sunlight (H=31.18, p

(Fig. 5.2b) .

Stolon spread The primary stolons (10) ofD. odorata branch at the nodes to produce secondary (2°) stolons and in this waythe plant increases occupation ofthe local environment. Significant differences for measured aspects ofstolon growth were found between plants grown under the different light treatments (p<0.05) (Table 5.2). By the end of 12 weeks the mean length ofthe primary stolons for plants in all treatments had increased by at least 3000 %, some measurements indicating that several plant stolons were growing up to 3 cm per day. Many plants produced secondary stolons, so that for some individuals the summed length of1° and 2 ° stolons exceeded eight metres. Both stolon length (1 °and 2°) and internode length increased with decreasing light availability (Table 5.2). When the mean internode length was determined for successive nodes along the primary stolon, the full extent of the differences in internode length became apparent (Fig. 5.3). All internode lengths were longer in the more heavily shaded clones than those receiving more light.

0 Although heavily shaded plants (17 % PAR) produced the longest 1 and 2 0 stolons, the proportion ofplants with secondary stolons and the number ofsecondary stolons produced per plant in this treatment was less than for those plants receiving more light (Table 5.2). Theoretical models predict an increase in branching frequency with an increase in light availability (Slade and Hutchings, 1987b; Solangaarachchi and Harper, 1987; Sultan and Bazzaz, 1993; Salonen, 1994). Contrary to expectation, branching in D. odorata was greatest at 45 % PAR, thus total stolon length was greatest for these plants largely as a consequence oftheir profuse branching (Table 5.2).

Dry weight production

Significant differences were found between the absolute dry weights of component plant parts (roots, leaves and stems) and the proportion of biomass allocated to these parts for plants grown at different light intensities (Fig. 5A a-d & Fig. 5.5). The dry weight of whole plants was greatest for those plants exposed to moderate light (45 %), followed by plants exposed to 71 % available sunlight and was lowest for those plants exposed to only 17 % PAR (Fig. 5Aa). These -76-

100 • 71 % PAR o 45 % PAR ""* * 17%PAR 80 EJ EJ * 0 * * * **EJ*EJ* ..-.. E [!]EJEJ EJ* * E ...... 60 * EJ [!] [!] EJ EJ EJ ..c Cl EJ[!][!][!][!]EJ -C ID o [!] [!] [!] [!] [!] * * .... tU [!][!][!][!][!][!][!] ID 40 • [!] [!] [!] [!] [!] • 0 • [!] [!] [!] [!] 0 [!] [!] [!] [!] • • [!][!][!]D 20 0 [!] • [!]

20 40 60 80 100 120 leaf width (mm)

Fig. 5.1. Mean length vs. width (mm) of 10 mature leaves of D. odorata plants exposed to one of three light levels, 71, 45 or 17 % PAR (n=60). -77- a

100 - * ** 80 - ~ ~ .I::.... 60 Q} - c ~ I I l Q) 0 40 I I I +J - I Q) o, 20 I - ~ I I I 71% 45% 17%

PAR

b

100

en Q) * >ro ....Q) 50 * 0 0 c ~ I 0

71% 45% 17%

PAR

Fig. 5.2. a & b. Petiole length (a) and leaf number (b) of D. odorata transplants after 12 weeks exposure to one of three light levels. Horizontal bar represents the median. Table 5.2. Summary ofthe results (means [±s.d.]); unless otherwise indicated) of measured characters of stolon spread for D. odorata plants grown at different light intensities for 12 weeks. n=60 plants measured for each character. Significantly different values are indicated with small letters (a.b,c); the difference between values with the same letter is not significant (p<0.05). Cl: confidence interval.

Treatment length (cm) internode length (cm) total stolon % plants with no 2° stolon/plant PAR (%) 1° stolon length (cm) 2° stolon length (cm) 2° stolon [95 % Cl]

17 141 [6.1]3 5.5 [0 .8]d 80 [32]& 173 [14.7] 36.6 0.5 [0.3-0.7] 45 122 [6.1]b 4.4 [0.9Y 74 [31]g 254 [30.4] 63.3 2.1 [1.6-2.6]

71 77 [4.1]C 3.4 [0.8Y 26 [13t 108 [8.8] 61.6 1.5 [1.1-1.9]

I ---.J 00 I -79-

6 ---E c -..c O'l -C ~ ID 4 "0 0 ....C ID -c

1 2 3 4 5 6 7 8 9 10 internode no. along stolon

Fig. 5.3. Mean length (cm) of the first 10 sequentially produced internodes along the primary stolons of D. odorata cuttings, following 12 weeks exposure to one of three light levels (n=60). Error bars represent standard error.

-71% -45% ----k-17% -80-

a: Whole plant biomass b: Leaf biomass

18 5 16 14 4 ::§12 :c 10 .~ 8 * ~ ~ ~ **' o--t.------r------r------,-- 0 4..--,-_--,-__..,---_ 71% 45% 17% 71% 45% 17% % PAR % PAR H=20.64 p<0.001 H=21.63 p<0.001

c: Stem biomass d: Root biornass

3 10- * -x- * ...... 2 .!:E *

* $ O~-,----,---..,---~ 71% 45% 17% 71% 45% 17% % PAR % PAR H=17.67 <0.001 H=27.13 <0.001

Fig. 5.4. a-d. Dry weights for whole D. odorete plants and component plant parts after 12 weeks exposure to different light levels (n=28). Horizontal bar represents median. -81-

80 • leaves Droots ~stems ~ 60 ------~ c: o ~ o ..Qm 40 en en co E o -:.a 20 --

o 71% 45% 17%

PAR

Fig. 5.5. Biomass allocation (%) to leaves, roots and stems by D. odorata cuttings, grown at 71, 45 or 17 % PAR (n=28). Error bars represent standard error.

~ 40 o o (j) o N 30 ------.------.------.-- o -"0 Q) co ...... ­g ~ 20 --(ij en en co E 10 o zs E Q) (j) 0 +--- 71% 45% 17%

PAR

Fig. 5.6. Stem biomass (%) allocated to secondary stolons by D. odorata cuttings grown at 71,45 or 17 % PAR (n=28). Error bars represent standard error. -82- results indicate that light intensity has an effect on average plant size. In terms ofthe magnitude of absolute weights, this pattern of response to available light (45 % > 71 % > 17 %) was observed for the absolute dry weights ofleaves, sterns and roots (Fig. 5.4. b-d).

The greatest percentage allocation ofdry weight by plants in alltreatments was to the sterns (Fig. 5.5). Furthermore, this proportion differed significantly between treatments and increased with decreasing light levels (F=15.7; p<0.05). The proportion of biomass allocated to secondary stolon production also differed and was significantly larger for those plants maintained at 45 % PAR (F=5.95, p<0.05) (Fig. 5.6). Proportional biomass allocated to the leaves was not significantly different at the range oflight intensities (F=I.26, p>0.05) despite the fact that there was a magnitude ofdifference in both the number and absolute dry weights ofthe leaves produced by the plants in each treatment (Fig. 5.2b & 5Ab). Allocation ofbiomass to roots for plants in all treatments was very low «20 %) and increased with increasing light levels (Fig. 5.5).

Flowers The flowering season for D. odorata is during the months ofMay-June. Forthis experiment very few individuals flowered and the experiment was terminated in late June when existing capitula were in full bloom. Although flowering was low for plants in all treatments, dry weight of flowers produced was greatest for plants grown at 45 % PAR (Fig. 5.7); 40 % (10 individuals) ofthe plants grown at 45 % PAR flowered while only 8 % (2 individuals) flowered at 71 % and 17% PAR. For this experiment flowering was not dependent on plant size as there was no relationship between plant size and production offlowers (Fig. 5.8).

5.4 Discussion

Delairea odorata responds strongly to a reduction in the availabilityoflight in terms ofits growth rate and resource allocation. As genetic variation was largely eliminated in our pot trial, the observed variation in plant performance can be attributed to phenotypic plasticity. After 12 weeks of growth, significant differences between plants, grown at different light levels, had developed in most measured variables. As far as absolute dry weight is concerned, the largest plants were consistently those exposed to 45 % available sunlight. Furthermore, plants grown at 45 % PAR produced the most leaves which were also larger and had longer petioles than -83-

1.6 ------.------.------._..~------.- -.------_ .. .. - -- ...-- ~ - .- -- -. --.

1.2

0.4

o -l----- 71% 45% 17%

PAR

Fig. 5.7. Total dry weight (g) of inflorescences produced by D. odorata transplants following eight months treatment at one of three light levels .

0.4 ---. Cl • • '-" (/) Q) 0 0.3 • c Q) 0 (/) ....Q) 0 ;;:: 0.2 c '+- 0 .c -Cl 0.1 • • 'iD • ~ • • • • .~ 0 • 0 5 10 15 20 25

plant weight (g)

Fig. 5.8. Dry weight (g) of inflorescences produced as a function of whole plant dry weight (g). T-test showed no significant difference between weight of non-flowering plants vs. flowering plants (p>O.OS). -84- leaves produced at 17 % and 71 % PAR. This suggests that the photosynthetic assimilation rate was highest at an intermediate light level; however, detailed physiological studies need to be conducted to confirm this observation. Zimdahl (1993) noted that weeds with great competitive ability are often shade tolerant in that their highest carbon dioxide assimilation does not occur in full sunlight. Mack (1996) suggested a viney growth habit and a low light compensation point

(the point at which the amount ofCO2 given out by the respiratory process is equal to that taken up by photosynthesis) to be traits indicative ofinvasive potential in natural or semi-natural forests. The most consistent plastic responses to availability oflight are manifested in internode elongation (etiolation) and branching (the proportion of axillary meristems which grow out) (Bazzaz and Harper, 1977; Sultan, 1987; Ellison and Niklas, 1988; Hutchings, 1988; Sutherland and Stillman, 1988; Hutchings and De Kroon, 1994; De Kroon and Hutchings, 1995). Most plants show reduced branching and internode elongation at lower flux densities (De Kroon and Hutchings, 1994), however, this is not always the case (Lovett Doust, 1987; see review by Hutchings and De Kroon, 1994). In an experiment to determine the effect oflight on growth ofthe clonal herb Glechoma hederacea L., Slade and Hutchings (1987b) found that clones grown under shading had significantly longer internodes than stolons produced by unshaded clones but total stolon length ofunshaded clones was over three times greater than for shaded clones, as a consequence ofprofuse branching. In this experiment, total stolon length, branching frequency and secondary stolon production was greatest for plants growing at an intermediate light level (45 % PAR) as opposed to those at 71 % available sunlight. Thus althoughD. odorata shows plasticity in growth response to light availability, it does not respond to changes in the light environment as predicted for plants with foraging behaviour. Instead resource acquisition appears to be consistently highest at 45 % PAR, a result consistent with observations on the success and proliferation ofD. odorata in moist, shady environments (Fagg, 1989; Chipping, 1993; Alvarez, 1997). Ifinternode lengths were shortest for plants exposed to 45% PAR then D. odorata would be foraging for intermediate light levels. However, internode length was consistently shorter at the highest light level (71 %).

The maximum weight ofinflorescences was produced by plants grown at 45 % PAR. However, as only a few plants flowered the results of this experiment should be interpreted with reservation. After eight months growth, there was no relationship between plant size and sexual reproduction, a relationship which has been reported in other studies (Whigham, 1974; Werner, -85-

1975; Pitkella et al., 1980). Pitkella et al. (1980) found sexual reproduction inAster acuminatus Michx., a forest understorey herb, to be indirectly affected by light through its effect on plant size. However, flowering is not always dependent on plant size (Salonen, 1994) and the results ofthis investigation indicate that flowering inD. odorata is independent ofplant size and directly affected by light availability. Further field tests need to be conducted to confirm this observation.

Finally, following data collection, it was realized that this experiment could be improved by subjecting the stolons of each replicate clone to different light regimes. As stolons grow predominantly horizontally, sets oframets along a stolon could be subjected, in experiments, to different light levels. Furthermore, to test for foraging behaviour, each set of ramets could be provided with a supply oflight, nutrients and water different from that supplied to other sets of ramets. Such an experiment would test implicitly the plasticity of a clone, as all variation in morphology could be ascribed to differences in growing conditions experienced by ramets ofan individual. Unfortunately there was insufficient time to conduct this experiment.

5.5 Concluding comments The results ofthis investigation raise some interesting questions regarding the growth pattern of Delairea odorata. However, further field and laboratory studies to determine the effect of resource availability(nutrient and moisture) on clonal developmentneed to be conducted before any conclusions regarding the growth strategy of this species can be made. Concurrent investigations into the habitat characteristics (interms ofresource availability)ofareas dominated by D. odorata should also be made. This study shows that (1) D. odorata can be easily propagated from stem fragments (as short as 3 cm) (2) D. odorata shows plasticity in growth form and deployment of biomass in response to light intensity (3) regardless of light level, greatest allocation ofbiomass is to stem growth (4) growth rate is highest at intermediate light levels, and (5) sexual reproduction is affected by light availability.Delairea odorata proliferates most vigorously in moist shady environments (Hilliard, 1977; Fagg, 1989; Chipping, 1993; Alvarez, 1997) and seldom grows in open habitats, such as grasslands, it is therefore not surprising that growth rate is highest at intermediate light levels. -86-

Chapter 6

The natural enemies of Delairea odorata and a potential biological control agent

Section 6.1. Natural enemies

6.1.1 Introduction Organisms which curtail plant growth or reproduction may be considered as potential biological control agents (DeBach, 1964; Hams, 1991; Shepherd, 1993). These include insects, mites, sheep, fungi, bacteria and viruses (DeBach, 1964). The search for natural enemies of a target weed should therefore encompass all organisms associated with the target plant. A high level of host-specificity is an essential attribute for any biological control agent. The first step towards biological weed control is thus the selection ofan host-specific organism from a range ofenemies (Harris,1991). For a classical biological programme host-specific natural enemies are collected in the country oforigin ofthe weed (Shepherd, 1993).

Nature ofthe controlling action The injury caused to a plant by an enemy may be either direct or indirect. Direct destruction pertains to natural enemies which destroy vital parts of the plant (e.g. roots, meristems and reproductive organs). An enemy may destroy a plant indirectly by facilitating infection by plant pathogens, thereby disrupting its competitive advantage (Anon, 1971;Bartlettand van den Bosch, 1971; Harris, 1973; Crawley, 1989; Harris, 1991). Since the response ofa plant species to tissue damage or infection varies according to survival strategy, the crucial stage to attack with a biocontrol agent also varies (Hams, 1991). Generally, much emphasis has been placed on the action ofinsects which attack seeds or flowers or which bore into roots or stems (Huffaker et a!., 1971; McClay, 1989; Muller, 1989; Crawley, 1989; McClay and Palmer, 1995). The point of importance is that whatever the nature ofthe injury, a good agent is one which through direct or indirect action causes the destruction of existing host stands, thus determining the latter's abundance; reciprocally its own abundance is then adjusted to that ofits host plant (Bartlett and van den Bosch, 1971; Andres et al., 1976; van Driesche and Bellows, 1996). -87-

Selection and evaluation ofnatural enemies Evaluation of natural enemies should comprise the following stages (Anon, 1971): 1. Regular surveys at local populations to obtain data on the organisms attacking the weed . 2. Concentrated studies on single control agents that cover life history, distribution, importance and enemies ofa potential control agent. 3. Host-specificity studies .

Surveys for potential control agents should be done over wide and ecologically diverse areas to maximise the chances offinding numerous natural enemies ofbroad genetic variability (Huffaker eta!. , 1971; Andresetal., 1976; Marohasy, 1989; Gilleteta!., 1991; McClay andPalmer, 1995). Plant species related to the weed should be included in the survey, as this provides preliminary information on the host-specificity of selected natural enemies (Muller, 1989; van Driesche and Bellows, 1996). In addition, surveys should be carried out in areas ofboth high and low weed density as this will enable identification ofcontrol agents with good searching ability (Bartlett and van den Bosch, 1971; Hams, 1973). The qualities ofan effective agent include: (1) an ability to reduce the competitive advantage of a weed (2) good dispersal and searching capabilities (3) adaptability to varying physical conditions (4) host-specificity, and (5) ability to Increase, or decrease, in number relative to the abundance ofthe host.

The present study Delairea odorata infestations are presently controlled by costly mechanical and chemical means (Cudney and Hodel, 1986; Fagg, 1989; Archbald, 1995; Bossard and Benefield, 1995; Moore, 1997; Forbert, 1998). Biological control may offer a cost effective alternative to the long term management ofD. odorata in exotic locations (Elliot, 1994; Balciunas pers. comm.). Elliot (1994) suggested the release oftwo insects introduced to control Senecio jacobaeae L. onto D. odorata. During 1994, experiments were conducted to test if these insects could reduce the vitality of D. odorata in the laboratory. Unfortunately, no follow up report on the results ofthese tests could be found. The alternative to using existing biocontrol agents is to search for the natural enemies ofD. odorata. This involves the development ofa classical biological control programme. Researchers from the USA'are ofthe opinion that in southern Africa the growth of D. odorata is stunted, possibly due to the action ofherbivores (Balciunas, pers. comm.). This has prompted surveys to determine if suitable biocontrol agents can be found in southern Africa. -88-

To date, only one other study ofthe insect fauna associated with D. odorata has been conducted

(Grobbelaar et aI., 1999).

The purpose ofthis study was to identify the natural enemies ofDelairea odorata and to assess the potential ofparticular species as biocontrol agents . To this end a survey was made ofthe fauna associated withD. odorata in KwaZulu-Natal. Preliminary studies were also conducted on the life cycle and host-specificity ofpotential biocontrol agents (6.2 Section 2).

6.1.2 Materials and methods FromFebruary to July and September to December (1998) surveys were made at regular intervals on populations ofD. odorata situated at Dargle, Ingele and Femcliffe (see Table 3.1 for location of sites). Plants at all sites were closely examined for damage by arthropods or other natural enemies. A minimum of one hour was spent collecting at each site on each occasion. The collected organisms were killed in jars charged with ethyl acetate. Later they were transferred to vials containing 70 % ethanol to preserve them for identification. Collections were biased as representatives ofgroups known to offer little or no prospect as biological control agents (such as crickets, grasshoppers and spiders) were not included. Any effect ofthese species on the rest ofthe insect fauna was not considered. During the flowering season, capitula were brought back to the laboratory and examined for the presence ofarthropods.

In field surveys it was not possible to determine if a captured insect was actually feeding on the plant. Published faunal lists for many plants frequently do not indicate whether the species are known to feed on the plant in question or were simply collected on it. This study reports all the phytophagous species which were collected on D. odorata; undoubtedly the list includes some species which are generalist herbivores/casual visitors . Most specimens were submitted for identification either to specialists within the School ofBotany and Zoology, University ofNatal, or to taxonomists at the Transvaal Natural History Museum or the Plant Protection Research Institute (pPRI). Most specimens have been identified at least to the level offamily.

6.1.3 Results and discussion

Many phytophagous species belonging to the insect orders Lepidoptera, Coleoptera, Hemiptera, and Diptera were found on D. odorata (Table 6.1). Most of the species are generalist Table 6.1. Phytophagous arthropods associated with Delairea odorata.

Order: Su perfamily: Genus & Species No. Adult! Frequency Collection Feeding (Suborder) F amily: Subfamily species Larvae (C/OIR) site type

Lepidoptera P yraloidea: Pyralidae Psara vercordalis Guenee 1 A C L Ect Ditrysia

Noctuoidea Noctuidae: Plusi inae Trichoplusia orichalcea F. 1 A 0 S (as pupa) Ect Noctuidae (larvae) not determined 4 LR L Ect

Arctiidae: Arctiinae Galtara purata Uzler 1 A R L Ect Arctiidae (larvae) not determined 5 L 0 L Ect

Geometriodea: Geometridae not determined 1 LC C Ect

P apilionidae: not determined 1 L R L Ect

Hemiptera Coreoidea: Coreidae not determined 2 A 0 L&S Ect Heteroptera Pyrrhocoroidea: Pyrrhocoreidae not determined 1 A C L&S Ect

Pentatomoidea Pentatomidae: Pentatominae not determined 4 A 0 L&S Ect

Lygaeoidea: Lygaeidae Nysi us sp. 1 AC C Ect

Homoptera Cercopoidea Cercopidae: Aphrophoridae spittle bugs 1 L C L&S Ect

Cicadelloidea: Cicadellidae not determined 3 A 0 L Ect

A: Adult L: larv ae Frequency: C: common 0: occasional R: rare I Collection site: L: leaves S: stems C: capitula 00 \0 Feedi ng type : Ect: ectophagous End: endophagous I Table 6.1. continued.

Order: Supe rfamily: Genus & Species No. species Adult/ Frequency Collection Feeding (Suborder) Family: Subfamily Larvae (C/O/R) site type

Coleoptera Chrysomeloidea Chrysomelidae: Alticinae Abrariussp. 1 A 0 C Ect Chrysomelidae: Chrysomelinae Ageniosa et badenii Vogel 1 A &L C L Ect Chrysomelidae: Galvercinae Prosmidia sp. 1 A R L Ect Chrysomelidae: Cryocerinae Bradylema sp. 1 A R L Ect

Scarabaeoidea Scarabeidae: Rutelinae monkey beetles 2 A 0 C Ect (Tribe: Hopliini)

Scarabacidae : Cetoniinae Leueoeellis sp. 1 A 0 C Ect

Cucujoidea Nitidulidae not determined 2 AC C Ect Phalacridae not determined 1 A &L C C Ect Cleroidea Melyridae: Mclyrin ae Pagurodaetylus sp. 3 A &L C C Ect

Curculionoidca Curculionidae not determined 1 A 0 C Ect Sciobius et bistrigieollis Boheman 1 A 0 L Ect Eremnus sp. 1 A 0 L Ect

Tencbrionoidca not determined 1 A&L C C Ect

Diptera Tephritoidea Cyclorrhapha Tephritidae Parafreu treta regalis Munro 1 A& L 0 S End

A: Adult L: larvae I Frequency :C: common 0 : occasional R: rare \0 0 Collection site: L: leaves S: stems C: capitula I Feeding type: Ec t: ectophagous End: endophagous -91-

(polyphagous) herbivores not suitable as biocontrol agents. Table 6.1 indicates the frequency of collection for each specimen. Because collections were not quantitative, the assessment of whether a given species is common, occasional or rare is based on a subjective assessment of collection frequency. Species listed as common were often found at collection sites in their geographic range. Rare species were collected only on one or a few occasions, usually as single individuals. Most species are ectophagous, feeding on leaves and/or stems.

Ten species were found associated with the capitula. Most ofthese are small beetles belonging to the families Melyridae, Nitidulidae, Phalacridae and Scarabaeidae (Subfamily: Cetoniinae and Rutelinae), families known to possess flower feeders . The scarabaeids and the nitidulids are generally polyphagous and therefore unsuitable biocontrol agents (Scott and Way, 1990). Internal feeding on the capitula by Phalacridae, Melyrinae and Tenebrionoidea larvae was observed. One lepidopteran belonging to the family Geometridae was found associated with the capitula. The larvae ofthis species are bright yellow and are frequently found spinning a loose cocoon on 2-3 capitula.

Only one leaffeeder may merit further investigation - a shiny black chrysomelid, Ageniosa cf badenii Vogel (Coleoptera: Chrysomelidae) (Table 6.1). Chrysomelids are generally host-specific and often completely defoliate their host. In addition, more species ofChrysomelidae have been used as weed biocontrol agents than any other family ofinsects (McEnvoy, 1984; lames et al., 1992; Julien, 1992; McEnvoy et al., 1993; Syrett et al., 1996). Although common, Ageniosa cf badenii was only collected at Ingele Forest. The adults were frequently observed mating on D. odorata and larvae were found feeding heavily on the undersurface ofIeaves.

One endophagous feeder was found - a stem galling tephritid fly, Parafreutreta regalis Munro (Diptera: Tephritidae) (Table 6.1). This species was found at all study sites and was identified as the natural enemy with the greatest potential as a control agent. -92-

Section 6.2 Studies on a potential biocontrol agent ­ Parafreutreta regalis Munro (Diptera: Tephritidae)

6.2.1 Introduction Biological control agents most frequently used against weeds ofthe Asteraceae belong to the family Tephritidae (see Chapter 2). Where tephritid stem-gallers have been used, outstanding control has been achieved. Procecidochares alani Steyskal and Procecidochares utilis Stone are credited with the control of Ageratina riparia R. King and H: Robinson and Ageratina adenophora (Sprengel) R. King and H. Robinson inHawaii, both serious economic weeds (Bess and Haramoto, 1959; Dodd, 1961; Harris, 1989; Julien,1992; Morin et al., 1997). The tephritids have been employed in biocontrol programmes largely for their host-specificity (Crawley, 1989). Most ofthe tephritid galls are formed on plants ofthe familyAsteraceae (Freidberg, 1984; Hams, 1989; Edwards et al., 1996). Plants belonging only to seven other families (Acanthaceae, Aquifoliaceae, Goodeniaceae, Melastomataceae, Mimosaceae, Onagraceae and Verbenaceae) have been reported to bear tephritid galls and less than ten species of flies are involved (Freidberg, 1984).

Before P. regalis can be considered a candidate control agent it is essential that studies be conducted into its biology. This study made preliminary observations, both in the field and the laboratory, on the life cycle and host-specificity ofthis species.

6.2.2 Materials and methods Field observations

In order to investigate the life cycle and host-specificity ofP. regalis, populations, situated across KwaZulu-Natal, ofD. odorata and its close relatives were frequently searched for mature galls. In southern Africa, closely related species to D. odorata includeMikaniopsis cissampelina, Senecio cinarescens, Senecio quinquelobus and (Jeffrey, 1986). These species were included in this study as a measure of the host-specificity ofP. regalis. Stems bearing mature galls were collected and brought back to the School of Botany and Zoology, at the University ofNatal, where they were potted and bagged with nylon-mesh to capture emerging insects. Data on the number of adults and pupae per gall and size of the gall at adult emergence -93- were recorded. Any specimens which emerged were collected for identification by Dr. R. Miller, University ofNatal, Department ofEntomology. Specimens are lodged at the Natal Museum.

Laboratory observations Further observations on the life cycle ofP. regalis were made in the laboratory in glass cages which contained, one or two, pottedD. odorata plants. The tops ofthe cages were covered with 1 mm nylon-mesh to allow both adequate air-circulation into the cage and occasional mist spraying ofthe potted plants. As soon as the young adults emerged from mature galls they were released into the cages and observations were made on their mating and oviposition behaviour. Following this, the plants were transferred to the greenhouse and monitored continuously. The number of galls which developed per plant and their rate of development was recorded. Furthermore, the time required for larval and pupal development was noted .

Light microscopy A light microscopy study ofstem anatomy ofD. odorata and the effects ofgalling thereon was carried out. This was done to provide some indication as to how P. regalis may be effective as a biocontrol agent. Thin transverse sections (TIS) of stems with and without galls were made for viewing with the light microscope. On galled stems sections were made ofthe tissue below, at and above the galls. All sections were stained with Toluidine Blue prior to viewing.

6.2.3 Results and discussion Field observations

Stem galls were found on only two of D. odorata's close relatives, S. tamoides and S. quinquelobus. The flies which emerged from these galls were Parafreutreta felina Munro (Diptera: Tephritidae) and Parafreutreta conferta Munro (Diptera: Tephritidae), respectively (Figs 6.1-6.3). Munro (1940) indicated that P. regalis, P. felina and P. conferta are allied in terms oftheir general biology and gall formation .

Figs 6.4 a-d show details ofthe galls collected from D. odorata, S. tamoides and S. quinquelobus. A total of60 galls were collected from populations of D. odorata. Ofthese only 20 (33 %) were successfully reared to adults; loss of gall forming maggots can be attributed to immaturity at collection, failure ofthe stem cutting to root (and therefore death ofthe gall) and internal fungal infection (of the gall) (Fig. 6.5). Of the three Parafreutreta species collected for this -94-

Fig. 6.1. Parafreutreta regalis (7.5X) Fig. 6.2. Parafreutreta conferta (7X)

Fig. 6.3. Parafreutreta fe lina (6.9X) -95-

50 ------.------.-- - -~ - --- . .- ~ - 4 ------.------..- -.- .-.- - --- o length • width

40 3.5 E §. Ql s: .~ 15>

o --\--.L-_ 2 conferta D. odorata S. tamoides S. qUinquelobus P. regalis P. felina P. species species

Fig.6.4a. Averagesize (length andwidth:mm) of Fig. 6.4b. Average length(mm)of pupae of P. galls. regalis. P.felina and P. conferta .

6 ------... --- 0 no. pupae • no. adults -- 60 5 --- -. ------..------.------

4 lii !:!! ~ 3 -- ­ E ::l C 2

0+--'--- P. regalis P. felina P. conferta P. regalis P. felina P. conferta

species species Fig. 6.4c. Average no pupaeandadultsper gall. Fig. 6.4d. % parasitised galls.

Figs 6.4 a-d. Data recorded form galls collected on D. odorata (n=20), S. tamoides (n=6) and S. quinquelobus (n=12) for the respective Parafreutreta spp. All values represent the mean, error bars indicate standard error. -96-

Fig. 6.5. Fungal infected galls (l2X). -97- investigation, the galls formed by P. regalis on D. odorata were the largest. They are soft, irregular swellings which may occur on the stem apex (acrocecidia), at a distance from the growing tip (up to 20 cm; pleurocecidia) or on the buds in the axils ofthe leaves. Up to eight adults emerged from a single gall. The galls formed byP.felina on S. tamoides are more elongate and woody than those ofD. odorata and were frequently found to contain fungal infections. The smallest galls were those formed by P. conferta on S. quinquelobus; no more than three adults emerged from a single gall. These galls were very woody and usually rounded in appearance. Generally, there seemed to be a definite tendency for larger galls to develop on younger, more succulent growth, typical ofthe fast growing stolons ofD. odorata. Ifthe plants are not growing quickly, the galls tend to be small with only one or two larvae.

Most species ofParafreutreta form "communal" galls where a few larvae develop together in the same cavity (Munro, 1940; Freidberg, 1984). In this study, a maximum ofnine puparia were recorded from a single gall on D. odorata. The puparia ofP. regalis are sub-translucent yellow and distinctly different from those ofP. felina, which are glossy opaque black (Figs 6.6 & 6.7). The puparia ofP. conferta are opaque yellow, the anterior end darkened, sometimes brown or almost black. According to Munro (1953) a curious aspect ofthe genus Parafreutreta is that some of the species seem to form pairs, where each ofa pair ofadults is very like the other but there is a marked difference in the shape and colour of their puparia; e.g. P. regalis and P. conferta.

Two species of parasitic Hymenoptera emerged from the P. regalis galls, both are species of Braconidae (Figs 6.8 a & b). These same species were collected from the P. conferta and P. felina galls on S. quinquelobus and S. tamoides, respectively. Up to 20 % ofthe P. regalis galls were parasitised. More galls will need to be collected to arrive at a more confident estimate ofthe percentage parasitism.

Laboratory investigations

In order to investigate aspects ofthe mating and oviposition behaviour ofP. regalis, young adults emerging from mature galls were transferred to glass cages containing one or two potted D. odorata plants. Soon after this, i.e. within five to ten minutes, mating was initiated. Before actually mating, however, both male and female engaged in a courtship dance where the male appeared to chase the female. Throughout this dance the adults continuously flexed their wings -98-

Fig. 6.6. Sub-translucent P. regalis puparia in the pith ofa gall (14X).

Fig. 6.7. Glossy black puparia typical ofP. felina (12.5X). -99-

a

b

Figs 6.8 a & b. Parasitic Hymenoptera (Braconidae) collected from D. odorata galls (7.5X). -100- and moved in semi-circular patterns. This ritualistic "rendezvous - behaviour" has been observed in other tephritid species and is associated with host plant recognition (Zwolfer, 1974; Freidberg, 1984; Harris, 1989). Meeting at the specific "rendezvous" point has been shown to be an important reproductive isolation mechanism for eight Urophora species (Tephritidae) that, deprived oftheir hosts, mate in various interspecific combinations which are rarely observed in the field (Zwolfer, 1974; Freidberg, 1984).

Following "rendezvous-behaviour" the flies mated for up to 30 minutes and within an hour the female oviposited on young, meristematic, apical shoots. The females frequently oviposited on all available growing tips, including axillary bud regions, for a period of10-20 minutes. On a few occasions they were observed to oviposit on mature leaves and stems. However, this was usually towards the end ofthe breeding term and no galls from these regions were recorded. Oviposition continued for up to 10 days after emergence, but generally terminated after one week. In the laboratory breeding trials, adult longevity was approximately 14 days (Table 6.2). A similar period for adult survival has been recorded for Procecidochares utilis Stone, also a stem galler, belonging to the family Tephritidae (Bess and Haramoto, 1959; Dodd, 1961; Bennett, 1986).

The eggs were preferentially deposited between the youngest pair of leaves at the tip of a vegetative stem or near the axillary buds. Unfortunately, the number ofeggs laid by the female on each apical tip was not recorded in this study. Early in the investigation it was hoped that it would be possible to rear enough flies to carry out more detailed life history and host-specificity studies; thus any possible source ofdamage to the young eggs, which may have been inflicted by counting under a light microscope, was avoided . However, due to time limits and difficulties experienced both in finding the galls in the field and in maintaining them in the greenhouse, it was not possible to continue with further investigations. Although not recorded in this study, larvae of gall forming Tephritidae usually hatch from eggs within several (3-8) days and the newly hatched larvae then enter the stem tissue (Freidberg, 1984).

Galls developed on all ofthe apical points ofthe potted D. odorata plants which were exposed to P. regalis. Only one or two galls developed on the axillary buds. Weekly measurements, made on the width and length ofthe developing galls, allowed data to be collected on the rate ofgall development and duration ofthe larval and pupal life stages (Figs 6.9 a & b; Table 6.2). In -101-

Table 6.2. Life history data collected for P. regalis. All measurements, except that of adult activity, based on data collected from galls initiated and raised to maturity in the laboratory (n= number ofgalls).

Stage in life cycle (n) No. days/months (±s.d.)

Tot. incubation period (15) 55 (19)

Larva (10) 30.6 (3.7)

Pupa (10) 17 (3.1)

Adult longevity (10) 14 (3.2)

Adult activity (60) September-June -102-

16

14

12

;--.. E E 10 '-" .c. Cl 8 -c:: Q) 6 «l Cl 4

2

0 1 2 3 4 5 6 7 time (weeks)

Fig. 6.9a. Weekly increase in length (mm) of developing 0. odorata stem galls.

10 ,------

8-1------+------

E E 6 +----- '-" .c. "0 - 1st "peep window" ~ 4+-- --- all"peep windows" present

2 +-----

o 1 2 3 4 5 6 7 time (weeks)

Fig. 6.9b. Weekly increase in width (mm) of developing D. odorata stem galls. -103- all cases the galls grew rapidly, some producing visible swellings within seven days ofoviposition. Maximum increase in gall size occurred two to three weeks after oviposition, thereafter growth decreased rapidly terminating during the sixth week. Other studies on the life cycle oftephritids have shown that the larvae commenced feeding almost immediately after hatching and continued until they were fully grown (Dodd, 1961; Freidberg, 1984). This is corroborated in P. regalis. Before pupating the larva prepared an exit for the young adult, sometimes referred to as a "peep window" or an "emergence window" (Bess and Haramoto, 1959; Freidberg, 1984). The larva scraped a certain spot in the wall ofthe gall, leaving only a thin layer through which the young adult can escape. ForP. regalis all "peep windows" were formed between weeks four and five, indicating that the larvae had entered into the pupal stage (Figs 6.9 a & b). Although a maximum offive "peep windows" were observed on a single gall, all adults emerged from a single window­ presumably made by the first larva to pupate (Fig. 6.10). Furthermore, for all ofthe mature galls collected in the field (n=55) only one peep window per gall was recorded and all adults emerged from this opening. Approximately two weeks after the appearance ofthe "peep-windows" the adults ofP. regalis emerged .

Parafreutreta regalis galls have been collected from D. odorata populations, at Dargle, Ingele and Ferncliffe, almost throughout the year, except during the very dry months ofJuly and August when the plants undergo severe die-back. The results from this study indicated the average time to adult emergence to be 55 days. However, a minimum period of 46 days (24/10/98-9/12/98) and a maximum 101 days (22/05/98-31/08/98) were recorded. Itis possible that temperature and time ofyear affected the duration ofthe life-cycle. Further data needs to be collected to confirm this observation.

Light microscopy : the anatomy of stems and the effects ofgalling thereon The anatomy ofyoung stems ofD. odorata is typical ofherbaceous dicotyledons (Fig. 6.11) (see Esau, 1960). The vascular system is eustelic. Individual vascular bundles are separated from one another but are mechanically strengthened by a band ofcollenchyma. There are approximately 12 vascular bundles in young stems.

Mature stems ofD. odorata develop secondary thickening. Secondary growth results from activity of the vascular cambium and increases the amount of vascular tissues in stems. It -104-

Fig. 6.10. The mature larva scrapes away a spot on the wall of the gall leaving a window through which the young adult can emerge. Generally only one "peep window" is formed in the P. regalis galls (lOX). -105-

Fig. 6.11. Transverse section (T/S) through a young D. odorata stem (l50X). C =band ofcollenchyma X= 10 xylem P=pith

Fig. 6.12. T/S ofa mature D. odorata stem showing "anomalous secondary thickening" (l75X).

NX=new xylem P=pith 2P=secondary parenchyma produced by the cambium -106-

contributes to the thickness of the stem axis but not to its length (Esau, 1960). In terms of secondary growth, vines, such as D. odorata, differ from non-climbing plants and are characterized by the presence of"anomalous secondary thickening" (Carlquist, 1991) (Fig. 6.12). In D. odorata, the vascular cambium originates in the fascicular and interfascicular regions (as with normal stems) and becomes continuous. However, the interfascicular cambium produces secondary parenchyma so that wide parenchymal rays are formed in continuity with the fascicular regions and the vascular tissues are separated into wedge-like blocks with secondary parenchyma in between (Fig. 6.12). New wedges ofvascular tissue may differentiate in the parenchymal rays. In D. odorata, young stems have approximately 12 vascular bundles while mature stems, with secondary growth, have up to 22 wedges ofvascular tissue. It has been hypothesised, and proved biomechanically, that the anomalous arrangement ofsecondary tissues in vines greatly increases the mechanical flexibility of stems and prevents breakage when stems are twisted or coiled (Carlquist, 1991).

The larvae ofP. regalis bore into the meristematic regions ofstems and an increase in stem girth occurs, i.e. "gall formation", as soon as the larvae begin feeding . Tephritid larvae generally feed by slashing at tissue with their mouth-hooks and eating the cell debris (Fig. 6.13) (Christenson and Foote, 1960). Increase in the size ofgalled stems is the result ofproliferation ofthe parenchymal tissue in the pith of the stem. During gall formation this tissue remains highly meristematic. Bennett (1986) noted that during the development of stem galls on Ageratina adenophora, formed by Procecidochares utilis, increase in volume ofthe pith region was often so great that there was lateral stretching ofthe cells. This was observed in the present study (Fig. 6.14) . Rapid multiplication and proliferation ofthe cells in the pith region has been associated with high levels ofcytokinins (Engelbrecht, 1971; Elzen, 1983; Lalonde and Shorthouse, 1984; Bennett; 1986). The larvae feed on the nutritive tissues ofthe extended pith until they pupate, at which stage cell division ceases (Bennett, 1986). Light microscopy ofgalled stems showed that as well as causing proliferation ofpith cells, galling also causes a disruption ofstem tissue below the gall (up to 3 cm) and adjacent to the gall (where galling occurs on axillary meristems) (Fig. 6.15 & Fig. 6.16). This tissue disruption may be associated with high levels ofendogenous cytokinins. -107-

Fig. 6.13. T/S ofa gall showing tephritid larva feeding on nutritive tissue. Note disruption of collenchyma tissue and vascular bundles (185X).

Fig. 6.14. T/S ofgalled stem tissue showing the highly meristematic pith region and lateral stretching ofcells (167X). -108-

Fig. 6.15. TIS ofstem tissue 3 cm below the gall (l70X).

Fig. 6.16. T/S ofstem adjacent to gall (when galling on axillary meristems) showing the disruptive effect of the gall (l65X). -109-

No stems with previous season 's galls were ever found in the field. It is possible that these stems suffer from indirect injury, through galling, and are weakened. This could be facilitated in a number ofways. Firstly, galling causes severe disruption ofthe pith parenchyma and, in addition to this, once the adult flies have emerged from the gall, a large central pith cavity remains. Disruption ofthe pith parenchyma may have adverse side affects on the surrounding conducting tissues. Ifthe xylem requires surrounding living tissues to act as an effective conduit then any large disruptions in the stem integrity may cause embolisms in the xylem resulting in water stress ofareas distal to that point. Furthermore, during secondary thickening, when there is an increase in conducting tissues, this effect would be compounded - especially as the secondary parenchyma (produced by the interfascicular cambium) cannot replace the pith parenchyma. Secondly, emergence ofthe adult flies leaves a broken epidermis on the surface ofthe gall. This provides a site for the entry ofother insect predators and fungal and bacterial pathogens which could kill the stem. ThusP. rega/is may reduce perennation ofD. odorata stems due to induced water stress and/or infection. Since no stems with previous season's galls were ever found in the field this seems highly probable.

6.2.4 Concluding comments

The genus Parafreutreta comprises 14 species all of which are cecidogenous on species of Senecio and their close relatives (Munro, 1952; Munro, 1953; Munro, 1957). Restriction of Parafreutreta to almost a single host genus, Senecio, as opposed to a tribe or family, indicates good specificity within this group, a valuable asset if this species is to be considered as a biocontrol agent. Furthermore, the genus Parafreutreta and its host plants are entirely restricted to south-central Africa (Munro, 1953; Freidberg, 1984). The lack ofclosely allied Senecio spp. in exotic locations may improve the specificity ofP. regalis in these regions (and decrease the likely occurrence offly predators). Ifthe distribution ofP. rega/is in southern Africa is limited to areas where D. odorata occurs, then this is also a good indication ofthe host-specificity ofthis species. Presently it is not possible to map the distribution ofP. regalis as only the type specimens (collected at Cedara, Natal) are lodged in the national collection at the Plant Protection Research Institute (Mansell, pers. comm.), and only four paratypes at the Natal Museum, Pietermaritzburg (Barraclough, pers. comm.) . -110-

Plant galls are active physiological sinks and can divert nutrients from other parts of a plant (Faucroy and Braun, 1967; Billet and Bumett, 1978; Bennett, 1986). The phenomenon ofgalls acting as nutrient sinks is an important component in the success achieved in programmes of biological control. Ifa gall can direct nutrients towards itself, or merely intercept nutrients from the translocation stream, then this could be to the detriment ofthe plant host. Bennett (1986) found that the galls ofProcecidochares utilis on Ageratinaadenophora acted as nutrient sinks, causing a significant reduction in flowering, achene production and vegetative growth. Other studies have recorded the adverse affects ofgalling on plant fitness (Bess and Haromoto, 1959; van Staden, Davey and Noel, 1977; Dennil, 1985). Based on the results ofthose studies, it can be predicted that, under heavy attack, galling of stems ofD. odorata by P. regalis is likely to have a detrimental effect in the form ofa reduction in sexual reproduction and vegetative growth.

This prediction needs to be empirically tested.

Lastly, although tephritids have frequently been employed as biological control agents against weeds of the Asteraceae, despite one or two outstanding examples, their overall success as control agents is debatable (see Crawley, 1989; Harris; 1989; Edwards et al., 1996). Poor establishment and lack ofsuccessful control by tephritids has partly been attributed to high rates of parasitisation in the new environment. It is often an expectation of biological control programmes that introduced insects will experience a lower level ofparasitism than in their native environments, and thereby have a greater impact on their host plant (Edwards et al., 1996). However, there remains the possibility that, in the new environment, native parasitoids will adapt to the novel host, and many examples exist to illustrate that this can occur (Comell and Hawkins, 1993; Edwards et al., 1996). Previous studies on parasitoids associated with tephritids have shown that ecological factors may be more important in determining host-parasitoid associations than phylogenetic factors, in particular the taxonomic relatedness ofthe tephritids (Hoffineister, 1992). Parasitism ofgall tephritids has been reported in the literature numerous times (Dodd, 1961; Freidberg, 1984) and is frequently associated with larval and pupal mortality - the most important factor regulating populations ofgall tephritids (Freidberg, 1984). The following are the most important families of parasitic wasps recorded from tephritid galls: Ichneumonidae; Braconidae; Cynipidae; Eulophidae; Eupelmidae; Eurytomidae; Pteromalidae and Torymidae (Freidberg, 1984). Only two species ofBraconidae were found to be associated with P. regalis. Ifthere are any more natural parasitoids ofthis species, this will need to be determined through -111- extensive collection ofgalls ofP. regalis from D. odorata across its southern African range. IfP. regalis is to be considered as a biocontrol agent predictions on potential parasitoids in California, Hawaii or Australia should be made. To do this a comprehensive sampling programme to survey the native parasitoids ofP. regalis in southern Africa must be carried out to determine the species involved and their relative importance. Furthermore, a sampling programme ofthe parasitoids of related tephritid species occurring in the region of introduction, e.g. California, must be carried out. Based on the current status ofknowledge oftephritid-parasitoid complexes, this information may give some indication as to the likelihood ofparasitisation occurring in the new environment. A similar method was used by Edwards et al. (1996) to predict the likelihood of parasitoids attacking three Mescolanis species, seed head tephritids of Chrysanthemoides moniliferaNorlindh, considered as potential biological control agents for this species in Australia. -112-

Chapter 7

Pyrrolizidine alkaloids

7.1 Introduction Biological activity ofpyrrolizidine alkaloids The association of plant secondary metabolism with plant-herbivore and plant-pathogen interactions is well documented (for review see Fritz and Simms, 1992; Romeo et al., 1995; Hartmann, 1996; Hartmann, 1999). Secondary metabolites have been found to render plants less susceptible to herbivore attack, acting as deterrents or antifeedants (Vrieling and van Vyk, 1994). The pyrrolizidine alkaloids (PAs), a class oftypical secondary compounds, are thought to serve as protective chemicals in plant-herbivore interactions (Molyneux and Ralphs, 1992; Hartmann et al., 1997; Hartmann, 1999). PAs are strong feeding deterrents for most herbivores; they are hepatoxic to vertebrates (Molyneux et al., 1979; Luthy et al., 1981; Johnson et al., 1985; Mattocks, 1986; Bicchi et al., 1989; Witte et al., 1993) and mutagenic to insects (Frei et al., 1992).

Although plants with PAs are usually avoided by generalist herbivores, a number of insect species from diverse taxa (for example, Lepidoptera, Orthoptera, Homoptera and Coleoptera) have evolved adaptations not only to cope with these compounds but also to use them to gain protection from enemies (Mattocks, 1986; Witte et al., 1993; Hartmann, 1999). These insects are usually host-specific and may use plant chemicals to locate or identify their host plants (van der Meijden, 1996). Their sexual behaviour may also be integrated in a variety ofways with host plant chemistry (for review see Landolt and Phillips, 1997). Specialist herbivores often meet, court and mate principally or exclusively on specific host plants (Landolt and Phillips, 1997). Certain insects sequester or otherwise acquire host-plant compounds and use them as sex pheromones or sex pheromone precursors. For example, certain butterflies ofthe subfamilyDanaidinae sequester PAs from larval food plants and, as adults, use them as defense substances or pheromones (Rothschild, 1973; Hartmann and Zimmer, 1986; Mattocks, 1986). Host plant chemistry thus plays an important role in the feeding and sexual communication ofa number ofphytophagous insects. Since PAs are frequently implicated in such interactions the PA profile of many plants is an important consideration in the determination of feeding patterns and host ranges of natural enemies, particularly that of specialist herbivores. -113-

Chemistry ofthe pyrrolizidine alkaloids The PAs encompass a diverse group of about 360 structures with a restricted occurrence in certain higher-plant taxa. They were first isolated from the genus Senecio (Compositae) but they have also been found in the Santalaceae, Boraginaceae, Rhizophoraceae, Poaceae, Orchidaceae and Fabaceae (Robins, 1982; Hartmann, 1996; Hartmann et aI., 1997). The PAs are ester alkaloids consisting ofa necine base (amino alcohol moiety), which is esterified to a necic acid (acid moiety) and are frequently accompanied in the plant by a variable proportion oftheir N­ oxides. They may occur as monoesters, open chain diesters or macrocyclic diesters (Mattocks, 1986; Witte et al., 1993). Esters of unsaturated necine bases (having a 1,2 double bond, e.g. retronecine) are referred to as "unsaturated PAs", whereas PAs with a saturated necine moeity (e.g. platynecine) are called "saturated PAs", even ifthere is unsaturation in the acid moiety (Mattocks, 1986) (see Base A & B). A plant does not usually contain a single PA, more often several alkaloids are present and these may vary considerably within a plant species with stage of growth, season and provenance (Borstel et aI., 1988; Witte et aI., 1993; van Dam and Frieling, 1994). Molyneux and Ralphs (1992) suggested the biosynthesis ofa number ofalkaloids to be a response to different insect enemies.

The present study One of the most important considerations in biological control is the host-specificity of the control agent as it is important to demonstrate that on introduction to a new area it will not damage plants ofeconomic or ecological importance (Huffaker, 1964; Zwolfer and Harris, 1971; Wapshere, 1974; van Driesche and Bellows, 1996). Since host-specificity in insects is largely dependent on host-plant morphology (e.g. presence of spines or burrs) and/or chemistry (e.g. alkaloids or essential oils), prior to any biological control programme it is equally important to investigate the degree to which the host plant is characterised by the presence of ecotypes or biotypes. Extensive intraspecific variation in the target weed can make biological control difficult as the ecotypes may differ in their susceptibility to the biocontrol agent (Andres et al., 1976; McClay, 1989; Shepherd, 1993). Furthermore, it may be difficult to identify effective biocontrol agents for a particular ecotype, and if several forms are present, several biotypes ofthe control agent may have to be introduced (McClay, 1989; Dennil and Hokkanen, 1990; van Dreische and Bellows, 1996). -114-

One ofthe aims ofthis study was to assess the suitability ofD. odorata to biocontrol. Since several studies have already been conducted on the PAs of this species (Adams and Gianturco,1956; Culvenor and Geissman, 1961; Stelljes and Seiber, 1990; Stelljes et aI., 1991) a study ofthe profile ofthese compounds for D. odorata across its native range will enable a comparison ofalkaloids to be made. IfD. odorata is characterised by a high degree ofvariation in alkaloid profile this may affect the complement of host-specific enemies associated with different populations. Furthermore, perhaps the breeding behaviour in exotic locations favours the formation ofgenetically isolated populations which may preserve specific or unique chemical characters. Ifso, this could have important implications for a biological control programme.

The most extensive investigation into the alkaloids ofD. odorata was made by Stelljes et al. (1991) on plants collected in California. Using GC-MS these authors detected the presence of 12 PAs inD. odorata (Table 7.1).

7.2 MaterialS and methods 7.2.1 Development of an analytical technique Before an analysis of the PAs of D. odorata could be undertaken, literature and laboratory investigations had to be carried out to determine a suitable extraction and isolation method for these compounds.

Extraction and isolation The PAs may be extracted either with methanol or dilute acid. Methanolic Soxhlet extraction is commonly used (Mattocks, 1967; Molyneux et al., 1979;Luthy etaI. , 1981; Johnson et aI., 1985; Hartmann and Zimmer, 1986; Witte et al., 1993; Krebs et al., 1996) and has been recommended as the first general procedure for complete extraction ofa PA source ofunknown composition. Typically, methanolic extraction is used for the extraction ofalkaloids from large samples (200 g-1 kg) (Koekemoer and Warren, 1951; Hartmann and Zimmer, 1986; Krebs et aI., 1996). For smaller samples « 10 g) and routine preparation ofmany samples, aqueous acidic extraction is recommended (Witte et al., 1993). As PAs occur both as their water soluble N-oxides and as tertiary bases, an excess ofzinc dust must be added to the acidic extract to convert all the N­ oxides into their GC-volatile tertiary bases. Following extraction, an Extrulet" column or a solvent partitioning method is used to isolate the alkaloids from the crude extract (Hartmann and Zimmer, 1986; Hartmann and Toppel, 1987; Stelljes and Seiber, 1991; Biller et aI., 1994; -115-

Table 7.1. Pyrrolizidine alkaloid (PA) profile for Delairea odorata as determined by GC-MS (Stelljes et al., 1991). Type refers to the alkaloid properties (type ofnecine base). R: retronecine P: platynecine M: monoester D: diester.

Alkaloid Type Percent of Total PA

anhydroplatynecine 6

7-tiglylretronecine RM trace amount

9-tiglylretronecine RM 22

7-angelylplatynecine PM 4

9-angelylplatynecine PM 39

7-tiglylplatynecine PM 4

9-tiglylplatynecine PM 8

sarracine PD 5

sarranicine PD 7

triangularicine RD 1

neosarracine PD 2

neosarranicine PD 4 -116-

Hartmann et al., 1997). The Extrulet" method is suitable for extraction and quantification ofPAs from small amounts ofplant material. However, it may not be quantitative ifvery polar PAs (free necine bases and esters with more than 2-3 free hydroxyl groups) are involved (Witte et ai., 1993). Partitioning against an organic solvent has been used by a number of investigators to isolate alkaloids (Hartmann and Zimmer, 1986; Stelljes and Seiber, 1990; Stelljes et ai., 1991; Witte et al., 1993). Following zinc reduction, the chlorophyll and waxes are removed by extraction of the aqueous acidic extract with petroleum benzene and/or diethyl ether. After basification of the acidic extract, the alkaloids are extracted into dichloromethane. The solvent is then removed and the residue resuspended in methanol.

Preliminary studies indicated acidic extraction combined with organic solvent partitioning to be the most reliable and practical methods with the highest yield ofalkaloids.

Separation and identification ofalkaloids Techniques for separation and qualitative and quantitative analysis ofthe PAs include: paper, gas and thin layer chromatography, high performance liquid chromatography (HPLC), gas chromatography-mass spectrornetry (GC-MS), spectrophotometry and carbon-13 NMR(nuclear magnetic resonance) spectroscopy (Chalmers et al., 1965; Mattocks, 1967; Molyneux et ai., 1979; Luthy et al., 1981; Mattocks, 1986; Roeder, 1990; Krebs et ai., 1996). Because ofthe variation in structural properties ofalkaloids, some techniques are better applied to some groups ofPA's than others. Spectrophotometric determination of alkaloids is a sensitive and reliable method but it is only effective for the identification and quantification ofPAs with an unsaturated necine moiety. Capillary gas chromatography combined with mass spectrometry (GC-MS) is the . most widely used technique for the analysis ofPA mixtures (Hartmann and Toppel, 1987; Biller et ai., 1994; Witte et al., 1993). When used in combination with the selective and sensitive nitrogen detector (NPD - Nitrogen Phosphorus specific Detector) GC-MS is a powerful high resolution technique (Hartmann and Zimmer, 1986). The retention indices (RI) in combination with [M]' and type specific MS fragmentation patterns can be used for identification of most PAs, including geometrical isomers.

Both NPD-detection and GC-MS were used for the identification of alkaloids. Furthermore, since Stelljes et al. (1991) had detected the presence of 12 alkaloids in D. odorata (collected in California) with GC-MS, it was assumed the present study would provide comparable data. -117-

7.2.2. Methods used for the extraction, isolation and identification of alkaloids Sample collection andpreparation To detect any differences in the alkaloid profile of local populations, plants from three locations across Natal (Harding, Dargle and Pietermaritzburg) were included in this study (plants were collected from two sites at each locality) (Table 3.1 lists the altitude and coordinates of each location). All plant material used in this analysis was collected in late May (1999) when D. odorata was in full bloom. To account for any diurnal variation in alkaloid accumulation plants were harvested between noon and 1:00 pm. The plant material, collected from each site, was separated into leaves, stems and flowers, air dried and then frozen (-18°C) until analyses started.

Alkaloid extraction and isolation The plant material (10 g dried leaves/stems/flowers) was ground in a Wareing Blender for 3-5 minutes in 100 rn1 0.5 N HCI, and extracted overnight with occasional stirring . After filtering the extract through glass wool , an excess of zinc dust was added to reduce all N-oxides to their corresponding tertiary bases. To allow for reduction, the mixture was stirred for 8-10 hours at room temperature. After re-filtering through glass wool, to remove zinc dust, 20 rnl of the aqueous acidic phase was extracted five times with 60 rn1 diethyl ether and subsequently

alkalinized with 5 rn1 saturated aqueous NaHC03. This basic aqueous phase was extracted four times with 60 rn1 dichloromethane . The solvent was evaporated under vacuum at 40°C and the residue redissolved in 3 rnl methanol.

Gas-Chromatography (GC) GC analysis was carried out using a Varian 3300 instrument fitted with a Nitrogen-Phosphorus specific Detector (NPD) and a 25 m, BP-5 bonded column (SGE, Australia) of0.53 mm internal diameter and lzzm film thickness. The injector temperature was set at 240°C and the detector at 300 °C. The attenuation ofthe detector was set at 4 x 10-12 mVN. Integration ofthe detector was by a Hewlett Packard 3395 integrator. The temperature programme for the analysis of samples was : 100 °C- l70 °C at 15°C/mill, 170 °C-250 °C at 8 °C/min followed by an 8 minute hold at 250°C.

Standard curve

The chromatograms were evaluated quantitatively using an external standard. Since pure samples ofthe alkaloids known to occur inD. odorata could not be obtained (limited funds) hyoscyamine -118-

(CUHDNO) 3 was used to construct a standard curve . Hyoscyamine contains a single nitrogen in the molecule as do the PAs, and the detector response is therefore comparable.

A standard curve ofdetector response vs concentration was prepared for a hyoscyamine dilution series. Six successive replicates of each concentration of the dilution series were analysed by injection of 1J-lI onto the GC. A straight line plot (regression analysis) was used to determine the relationship ofpeak area to concentration (Appendix A. Fig. 1).

Gas ChromatographylMass spectrometry (GC-MS) GC-MS identification ofthe alkaloids was carried out with an HP 5988A instrument fitted with a 15 m, HP-I column of 0.25 mm internal diameter and Izzm film thickness. The injector temperature was set at 220 QC. The temperature programme for the analysis was: 40 QC-270 QC at 10 "Czmin.

7.3 Results and Discussion t The alkaloids ofDelairea odorata Gas chromatography ofthe purified alkaloid extracts ofD. odorata revealed the presence ofat least nine possible nitrogenous compounds (Fig. 7.1). GC-MS indicated the presence of nine pyrrolizidine alkaloids (Fig. 7.2, Table 7.2). However, as both the structure and polarity ofthe column on the GC (BP-5) were different from that on the GC-MS (HP-I) and instrument sensitivity differed, most ofthe pyrrolizidine alkaloid peaks identified by GC-MS could not be correlated with the nine peaks obtained on the GC with the NPD-detector. Only the most prominent GC-peaks, GC peak 2 (10.883 min) and GC peak 3 (11.106 min) (Fig. 7.1), could be tentatively identified as GC-MS peak 4 (9.56 min) and GC-MS peak 5 (9.61 min) (Fig. 7.2, Table 7.2; for review on problems experienced with peak resolution when using different column types see Witte et a!' (1993)).

When analysing mass spectral data the fragmentation pattern unequivocally helps to distinguish between esters ofunsaturated necine bases (e.g. retronecine) and esters ofsaturated necine bases

7 9 (e.g. platynecine) as well as 0 _ and 0 _ monoesters (e.g. 7-angelylplatynecine and 9-angelyl­ platynecine) (Mattocks, 1986; Witte et a!., 1993). Monoesters of saturated necines, such as platynecine, typically have an [M]' 239 and give fragments in the ranges m1z 95-97, 122-123 and 138-140; there is a characteristic base peak at m1z 82 or m1z 95 for the 0 1 (e.g. 7-tiglyl- ~'J

~I ~ I -l i eo I cr': ...... p­ er, U )

I b~1 31 ...... ':j HI' I li i 11 ! i I! 'I ' n r- I P. . u !l ~ c 11 ',' ,

C- . i ·.J III \I I I Lf') j"-l ,=, 1:::9 lj ~ ,I '~~ 1 . I'

. ,il \11 i \}\", I, 11 t.JI ,\ J! j 1.1 '.' '... I., ., . ,', ~ '., L, l _.1 , .1. ;,·y::.:-:-"=i:-:.::..·..:;'i..:..-..:.' .!...._· · { ·~~ :~~,+ · ..----r - -_ · _- ., ·_·~~ ....: :~~ :: : ~ :: · : ~ :: ::.: ~ - · - l ..:.. - ~ '- .., · · :. · ~ ::: :; :·::::.::: : l "------~~"'-' 8 10 12 I Time (minutes) Time (minutes)

I

-\0 -I Fig. 7.1. Separation ofan alkaloid extract fromDelairea odorata by GC. Fig. 7.2. GC-MS ofalkaloid extract. Table 7.2. Mass spectra ofpyrrolizidine alkaloids extracted from Delairea odorata collected in Natal. ([ ] indicate relative abundance ofions)

Other characteristlc base ions

Peal< Rt [M]+ 1 2 3 4 5 6 7 8 9 10 11' 12 13 14 15

1 9.01 237 80 111 93 106 94 137 136 138 81 124 120 108 71 83 237 [100] [49] [44] [40] [34] [29] [18] [15] [14] [14] [11] [10] [10] [9] [5]

2 9.07 237 93 138 94 80 137 136 111 106 155 193 124 108 237 154 [100] [53] [38] [32] [24] [14] [12] [8] [5] [3] [4] [4] [3] [2]

3 9.25 237 80 106 111 137 94 124 136 81 83 237 154 [100] [43] [36] [29] [27] [24] [18] [13] [12] [3] [2]

4 9.56 239 80 93 137 106 94 111 83 124 136 138 154 155 193 [100] [78] [66] [59] [49] [47] [37] [35] [35] [22] [18] [12] [3]

5 9.61 239 93 137 94 138 83 136 155 154 80 108 106 124 193 [100] [33] [25] [23] [19] [18] [17] [17] [17] [4] [3] [2] [4]

6 9.85 237 93 149 109 81 127 139 94 71 80 69 95 83 136 106 111 [100] [90] [84] [68] [60] [56] [39] [42] [46] [39] [37] [34] [31] [20] [19]

7 9.87 237 93 136 83 94 80 119 137 120 106 81 111 118 154 117 [100] [68] [52] [51] [35] [26] [21] [16] [14] [10] [8] [8] [8] [8]

8 10.5 285 93 193 83 94 80 194 67 92 285 106 82 [100] [63] [16] [14] [9] [8] [6] [4] [4] [3] [3]

9 11.9 251 220 136 93 83 120 94 119 221 80 251 121 137 118 106 [100] [58] [41] [37] [32] [29] [20] [17] [15] [10] [9] [9] [9] [8]

...... I N o I -121- platynecine) or 0 9 (e.g. 9-tiglylplatynecine) esters respectively (Mattocks, 1986; Witte et al., 1993). Esters ofunsaturated necines have an [M]' 237, the 07- esters (e.g. 7-angelylretronecine) show key fragments at rnlz 80, 106 and 111 and the respective 09-esters (e.g. 9­ angelylretronecine) fragment at rnlz 93, 136,137 and 138 (Witte et aI., 1993).

Using plants collected in California, Stelljes et al. (1991) found 12 pyrrolizidine alkaloids inD. odorata (Table 7.1). Table 7.3 shows the mass spectral data ofall the alkaloids isolated by these authors. The major alkaloid was 9-angelylplatynecine which comprised 39 % ofthe total alkaloid content (Table 7.1). 9-Tiglylretronecine was also present in large amounts but was mixed with lesser amounts of7-tiglylretronecine (Table 7.1). Thus both saturated (platynecine based) and unsaturated (retronecine based) PAs were present in the material analysed by Stelljes et al. (1991), although the saturated PAs were by far the most abundant.

The fragmentation patterns of the GC-MS peaks of material analysed in this study did not correlate with that of Stelljes et al. (1991) (compare Tables 7.2 & 7.3). Most of the alkaloids identified by those authors were platynecine mono esters or diesters with a characteristic base peak at rnlz 82/95 (Table 7.3). With the exception ofthe alkaloid identified at peak 9, all ofthe alkaloids identified in this study have a base ion ofrnlz 80/93 (Table 7.2). This indicates that most ofthe alkaloids present in material analysed for this study are unsaturated retronecine (07and 0 9) monoesters. Key fragments at 220, 93-95, 119-121 and 136 are all characteristic ofretronecine diesters (Toppel et aI., 1987), fragments obtained for GC-MS peak 9 correspond to this. No platynecine based (saturated) alkaloids were present in D. odorata collected in Natal.

GC-MS peaks 4 and 5 have a [M]' 239, and a base peak at rnlz 80 and 93 respectively (Table 7.2). This is unusual as the base peak fragments (and others) are key fragments ofretronecine mono esters yet characteristically retronecine esters have an [M]' 237. This indicates that for peaks 4 and 5 there has either been incomplete separation and two compounds are eluting as a single peak, or that unknown alkaloids are present in the plant material. Only one pair of unsaturated monoesters was found by Stelljes et aI. (1991),7 and 9-tiglylretronecine (Table 7.1 & Table 7.3), the angelate esters ofretronecine were not detected in their extract. Interestingly, the fragmentation pattern for GC-MS peak 2 corresponds exactly to that of9-angelylretronecine published by the same authors for Senecio hydrophilus Nutt (Table 7.4). This may indicate the presence ofthese isomers in locally distributed plants. Table 7.3. Mass spectra of pyrrolizidine alkaloids extracted from Delairea odorata by Stelljes et al. (1991). ( [ ] indicates relative abundance ofions)

Other characteristic base ions

Alkaloid Rt [M]+ 1 2 3 4 5 6 7 6 8 9 10

anhydroplatynecine 8.79 139 82 139 138 96 110 95 120 122 [100] [29] [34] [15] [9] [9] [5] [4]

7-tigl ylretronecine 23.24 237 80 93 94 106 137 136 III 138 124 154 108 [100] [61] [44] [34] [32] [23] [23] [20] [17] [9] [7]

9-tiglylret ronecine 23.30 237 93 80 94 138 137 136 106 154 124 III 108 [100] [49] [43] [37] [36] [20] [15] [17] [10] [9] [9]

7-angelylplatynecine 24.37 239 82 139 83 156 140 138 114 122 95 239 96 [100] [59] [48] [38] [37] [20] [12] [8] [7] [7] [6]

9-angelylplatynecine 25.71 239 82 95 96 122 83 221 140 239 138 139 156 [100] [99] [46] [14] [13] [11] [9] [6]- [5] [4] [2]

7-tiglylplatynecine 26.16 239 82 139 156 138 83 95 114 96 140 122 239 [100] [53] [27] [22] [20] [12] [11] [8] [6] [5] [1]

9-tiglylplat ynecine 27.68 239 95 82 221 96 83 122 140 138 139 156 [100] [74] [15] [14] [11] [6] [5] [4] [3] [1]

sarracine 48.59 337 138 82 122 83 96 139 123 237 222 121 97 [100] [65] [54] [38] [37] [35] [29] [20] [17] [15] [9]

sarranicine 49.25 337 138 82 95 122 139 96 123 83 121 237 222 [100] [58] [43] [41] [30] [29] [27] [26] [16] [16] [12]

triangularicine 49.68 335 83 93 136 94 120 119 80 121 81 220 137 [100] [68] [66] [64] [46] [37] [35] [33] [23] [21] [16]

neosarracine 50.05 337 138 82 122 95 96 139 83 123 222 237 121 I [100] [78] [62] [45] [42] [38] [37] [34] [18] [18] [17] ...... N N neosarranicine 50.68 337 138 82 95 122 139 123 96 83 121 237 222 I [100] [76] [56] [53] [38] [36] [36] [34] [21] [19] [17] -123-

Table 7.4. Mass spectral data for 9-angelylretronecine (Stelljes et al., 1991).

Alkaloid [MJ+ 1 2 3 4 5 6 7 8 9 10 11

9-angelylretronecine 237 93 138 94 80 137 136 III 106 124 108 154 relative abundance [lOOJ [49] [46J [44] [36J [15J [10J [10] [9] [6] [3]

It was expected that similar compounds to that isolated by Stelljes et al. (1991) would be present in plants collected in Natal and that GC-MS fragmentation patterns would enable identification ofalkaloids. However, obviously different fragmentation patterns were obtained. Based on the fragmentation patterns alone none of the alkaloids obtained for this study can be positively identified. To do this , more analytical data needs to be collected - this includes isolation of individual alkaloids and an HPLC analysis. Nuclear magnetic resonance (NMR) can be used to identify unsaturated PAs and when used in combination with GC-MS is an effective technique to elucidate structural isomers (Hartmann and Zimmer, 1986). Many ofthe alkaloids in D. odorata (in Natal) are retronecine esters, a NMR analysis may help to identify the individual alkaloids. Since it was anticipated that known alkaloids would be found an exhaustive analysis to identify compounds falls out ofthe scope ofthis study. All purified alkaloid samples have therefore been pooled and this sample will be used to positively identify alkaloids at a later stage.

As well as providing information on the types ofalkaloids inD. odorata collected from Natal, this study also examined the alkaloid profiles oflocally distributed plants. To this end plants were collected from two sites at each of three locations across Natal. The distribution of alkaloids within the plant was also investigated. Prior to analysis plants were separated into flowers, leaves and stems.

Distribution of alkaloids within plants and between populations The same alkaloids were extracted from plants collected at Ingele, Ferncliffe and Dargle (Fig. 7.3 a-f). For all plants, and their component parts, the most abundant alkaloid was that occurring at GC-peak 2 (10.883 min, Fig. 7.1)(a retronecine) (Fig. 7.3 a-f). This alkaloid comprised at least 70 % of the total alkaloid content of all samples. At all sites, the flower heads contained the highest concentration ofalkaloids and more than 80 % ofthe total alkaloids ofplants in full bloom were present in the capitula (Fig. 7.3 a-f& Fig. 7.4) . Much lower concentrations ofalkaloids were present in both the stems and leaves (Figs 7.3 a-f& Fig. 7.4). Hartmann and Zimmer (1986) and Borstel et al. (1988) found PAs in Senecio vulgaris L. and S. vernalis L. to accumulate mainly in the inflorescences. Hartmann and Zimmer (1986) found concentrations of -124-

a. Ingele: Site 1 b. Ingele: Site 2 1000 1000

100 100

10 - 10

T [- , ~ ~ ~ , ~ ~ 0.1 -r '-r 0.1 2 3 4 5 6 7 8 9 2 3 456 7 8 9

1000 c Dargle· Site 1 d. Dargle: Site 2 1000 c ca 100 +----J_._------100 +--,..- f------...C) o c 10 r-- 10 ~....__---tlI::t I---~,---__III------ca C­ C) 1- C) -::J

2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9

e. Ferncliffe: Site 1 f. Ferncliffe: Site 2 10000 10000 - 1000 1000 +-- w_------

100 - - 100 ~--_ :1-__- __------

10 I- 10 :

~ ~~ ! ~ ~~ 0.1 n 0.1 2 3 4 5 6 7 8 9 2 3 456 7 8 9

GC peak no.

Fig. 7.3 a-f. Pyrrolizidine alkaloids (ug/g) found in flowers, leaves and stems of D. odorata plants collected at two sampling sites at 3 localities ; Ingele (a&b), Dargle (c&d) and Ferncliffe (e&f). flowers • leaves lE stems 0 -125-

1

0.8 • stems

"'C o leaves ·0 0.6 -- l;.'=J flowers ~ro co ci. 0 '- 0.4 0-

0.2

o Ingele 1 Ingele 2 Fern 1 Fern 2 Dargle 1 Dargle 2

location/site no.

Fig. 7.4. Relative distribution of pyrrolizidine alkaloids within different D. odorata plant parts. Plants collected from two sampling sites at three localities: Ingele, Femcliffe and Dargle. -126- alkaloids in mature flower heads ofS. vernalis to exceed that ofvegetative organs 5-10 fold and to account for up to 80 % oftotal plant alkaloids.

Plants collected from Femcliffe had by far the highest concentration of alkaloids (Fig. 7.5 a-d). Therefore, although alkaloid profiles ofplants collected from the three populations were similar, in terms ofconcentration ofalkaloid, plants collected at Femcliffe had more alkaloid than those at Dargle and Ingele. Variation in alkaloid concentration between populations, and even individual plants, is not uncommon (Johnso n et al., 1985; van Dam and Vrieling, 1994).

7.4 Concluding comments Two basic types of alkaloids can be distinguished from the plant material collected in KwaZuluINatal - retronecine monoesters and one diester. The results differ markedly with previous studies which have found principally saturated alkaloids, either sarracine (platynecine diester) (Culvenor and Geissman, 1961; Stelljes and Seiber, 1990) or 9-angelylplatynecine (monoester), to be the major alkaloids. It is interesting to note that a repeated study by the same authors (Stelljes and Seiber, 1990; Stelljies et al., 1991) yielded a different major alkaloid. Stelljes et al. (1991) suggest isomerisation of sarracine to neosarracine, sarranicine and neosarranicine during sample storage to account for some ofthe variation observed in sarracine content from 1990 to 1991. Sarracine is an apparently rare alkaloid reported only in D. odorata and Senecio atratus Greene (Stelljes and Seiber, 1990), the absence ofthis alkaloid from the material analysed in this study is therefore noteworthy.

The results ofthis study suggest that the alkaloid profile oflocally distributed plants differs from those occurring in California and possibly even Australia (Culvenor and Geissman, 1961; Stelljes and Seiber, 1990; Stelljes et a!., 1991). The fact that the same alkaloids, with similar profiles, were extracted from plants distributed across KwaZulu-Natal serves to reinforce the apparent difference in alkaloid profile between local plants and those in California. These results indicate that chemotypes ofD. odorata may be present. However, a more exhaustive comparative analysis including seasonal material collected from throughout South Africa and from Australia and California needs to be conducted to confirm these findings . Furthermore, this data should be collected prior to any biological control initiatives.

Pyrrolizidine alkaloids are present in all parts ofD. odorata, however, there is a conspicuous accumulation ofalkaloids in the flower heads . As already noted, the pyrrolizidine alkaloids play -127-

a. Total weight b. Capitula 30 25

25 20

20 15

15

10 10

5 5

"t:J .-0 0 0 co Ing 1 Ing 2 Fern 1 Fern 2 Darg 1 Darg 2 Ing 1 Ing 2 Fern 1 Fern 2 Darg 1 Darg 2 ~ CO C) E 3.5 c. Leaves 1.4 d. Stems

3 1.2

2.5

2 0.8

1.5 0.6

0.4

0.5 0.2

0 0 Ing 1 Ing 2 Fern 1 Fern 2 Darg 1 Darg 2 Ing 1 Ing 2 Fern 1 Fern 2 Darg 1 Darg 2

location/site no.

Fig. 7.5 a-d. Weight (mg/1 0 9 dry plant material) of pyrrolizidine alkaloid extracted from (a) whole plant (leaves, flowers and stems) (b) capitula (mg/10 g) (c) leaves (mg/10 g), and (d) stems (mg/10 g) of D. odorata, collected at two sites for three geographical localities: Ingele, Ferncliffe and Dargle. -128- an important part in plant-insect interactions. The predominance of alkaloid accumulation in flowers may simply be interpreted as effective protection ofa very valuable plant organ against herbivory. -129-

Chapter 8 Summary and conclusions

Applicability ofbiological control to any weed problem depends on the nature ofthe weed, its taxonomic position, the degree of control required, the availability ofsuitable enemies and the stability and complexity of the plant community which is invaded (Andres et al., 1976; van Driesche and Bellows, 1996). Generally, successful biological control has occurred mainly in families containing few plants of economic or ecological importance (Andres et al., 1976; Shepherd, 1993; van Driesche and Bellows, 1996). Mode ofreproduction is also an important determinant of the success or failure ofa biological control programme. Burdon and Marshall (1981) noted the greater success ofbiological control programmes against asexual weeds.

Delairea odorata is an important weed ofmany nature conservation areas (Fagg, 1989; Scott and Delfosse, 1992; Alvarez and Cushman, 1997) and biological control is a potential method for the management ofthis species in exotic locations. As Delairea is a monotypic genus, no closely related crop or ecologically important species, which may be targeted by introduced enemies, have yet been identified. This makes biological control ofthis species a viable option. For this investigation emphasis was placed on the reproductive biology, the natural enemies and the chemical ecology of D. odorata. All of these attributes were recognised as key aspects in determining the suitability ofD. odorata to biological control.

The surveys for natural enemies ofD. odorata, in KwaZulu-Natal, indicate that this species has a wide range ofenemies, although many appear to be generalist herbivores. One potential control agent was identified, a stem galling tephritid fly, P. regalis. The preliminary observations made in this study indicate this species to be host-specific, a valuable asset if it is to be used in a biocontrol programme. Furthermore, as D. odorata proliferates extensively by means of stem regeneration and elongation, galling ofthese growing points by P. regalis may limit stolon spread in exotic locations. Further studies to determine the potential of P. regalis as a biocontrol agent must, however, be carried out. These should include controlled greenhouse and field tests designed to evaluate (1) fly population growth (2) the effect ofgalling on stem regeneration, and (3) effect ofgalling on sexual reproductive effort. Conclusions regarding the limiting effect of galling on plant population spread and dispersal may then be drawn and the suitability ofP. regalis as a biocontrol agent determined. -130-

The pyrrolizidine alkaloid analysis shows that the same alkaloids are present in populations ofD. odorata scattered across KwaZulu-Natal. However, these alkaloids appear to be different from those occurring in plants found in California and possibly even Australia (Culvenor and Geissman, 1961; Stelljes and Seiber, 1990; Stelljes et al., 1991). This is an interesting and important result which indicates that chemotypes ofD. odorata may be present, a factor which may affect the likelihood of natural control agents establishing on exotic plant populations. Further studies to confirm these preliminary findings are therefore essential. For veritable interspecific populational comparisons to be made, follow up investigations should include an analysis ofthe pyrrolizidine alkaloids of plants collected from throughout South Africa, Hawaii, California and South Australia. These resultswill indicate, conclusively, whether chemotypes ofD. odorata are present or not. An examination ofthe seasonal and ontogenetic change in alkaloid profiles should also be made. A record ofseasonal change (ifany) in alkaloid profiles will indicate at what time ofyear the plants are most susceptible to enemies as this may be important regarding the timing of release ofspecific control agents in exotic locations. Furthermore, juvenile plants may be found to contain high proportions ofalkaloids, as compared to adults, protecting them from enemies. This information will be useful to any control initiative.

This investigation has shown that D. odorata is a shade tolerant, self-incompatible species which reproduces sexually by seeds and asexually by stolons . Typically, the most suitable breeding system for colonising new areas is "self-compatible" (Baker, 1965). Selfing provides obvious benefits to a colonist allowing it to reproduce when mates are absent or at low density, as is often the case during the initial stages of colonisation. An association in plants between colonising success and uniparental reproduction has frequently been reported (Barret and Richardson, 1986; Abbot and Forbes, 1993). However, as is the case with D. odorata, not all successful colonisers are capable of uniparental reproduction and this raises interesting questions as to how they overcome the constraints imposed by the requirements for mates during the initial stages of colonisation. Baker (1974) suggested that for perennial weeds, where self-compatibility is less certain to be found, prolific vegetative reproduction is often present and this achieves the same end, i.e. the rapid multiplication ofindividuals with appropriate genotypes. This observation was supported by Vogt Anderson (1995) who found vegetative reproduction to be more important than sexual reproduction for invaders ofnatural habitats . D. odorata reproduces extensively by stolons which have the potenti al to produce a new plant at each node. In addition, thick rhizomes -131- support regrowth ofaerial shoots when they are damaged or removed. Vegetative reproduction thus plays an important role inthe local spread ofthis species and has facilitated invasion in exotic locations. One other possible answer to the question of overcoming mate requirements during the initial stages of colonisation was suggested by Abbot and Forbes (for perennial species; 1993). Following arrival at a new site perennials will survive to flower for more than one year. This population of colonising individuals could expand in numbers and become established via seed immigration over two years or more before engaging in sexual reproduction. However, this theory requires colonisation by more than one individual.

In California, no viable seed has yet been collected from D. odorata (Balciunas, pers comm.; Alvarez, 1997). This may indicate that most ofthe plants in California spread vegetatively by stolons and that the lack of seed set is due to- clonal populations. Pollination experiments to determine whether geographically separated Californian populations have the potential to reproduce by seed are presently being carried out (Robison, 1999). If, as present information suggests, the Californian populations are largely clonal, this has a number of important implications with regard to the biological control of D. odorata in this region. Asexual weeds with genetic homogeneity may potentially be controlled by a single agent; a valuable asset if biological control is to be attempted. However, it is important to remember that interpopulation variation in a plant species may arise due reproductive isolation - be that geographic, seasonal or temporal (Housard and Escarre, 1995). Due to reproductive isolation exotic populations ofD. odorata may be distinct ecotypes of their native counterparts. Although further testing is required, results ofthe pyrrolizidine alkaloid analysis indicate that chemotypes may be present. This could present problems in the development ofa control programme.

In Australia and Hawaii, D. odorata apparently spreads both vegetatively and by wind blown diaspores (Lawrence, 1985; Jacobi and Warshauer, 1992; Cox, 1998). As previously mentioned, the nature ofthe contrasting modes ofreproduction ofD. odorata in exotic locations raises some interesting questions as to the genetic attributes and the nature ofthe spread ofpopulations in each exotic location. This will influence the measures taken for its control in each of these regions. Natural enemy surveys carried out for this study revealed several species associated with the capitula ofD. odorata. Although most appear to be generalist herbivores further, and more detailed, studies on some ofthese enemies may yield useful control agents . A good control agent -132- is one which targets the life strategy ofthe host plant. Furthermore, good control of weeds ofthe Asteraceae has been achieved when multiple enemy introductions have been made (e.g. control programme against Senecio jacobaeae). Introducing both seed and shoot enemies may limit the spread ofD. odorata in these regions .

Although numerous studies have been conducted to characterise attributes common to invasive species (Baker, 1974; Lawrence, 1980; Rejmanek, 1989; Rejmanek et al., 1991; Perrins et al., 1992; Vogt Anderson, 1995; Pysek, et al., 1995; Rejmanek, 1995; Mack, 1996; Williamson and Fitter, 1996; Vermeij, 1996) as yet no list ofdefining characters has been published. Amongst a suite of other characters, climate (homoclines), habitat requirements and absence of natural enemies have been cited as important factors regulating the spread ofplant populations in new environments. South Africa is overall an extremely dry country with sporadic moist patches. Here, D. odorata has a limited distribution and occurs only in scarp forests 800 m above sea level, which are dominated by cool climate with high annual rainfall (the mist-belt region). In southern Africa the spread ofD. odorata may be limited to scattered wet areas, hence its patchy distribution . In South Australia, California and Chile, D. odorata occurs along the coastal belt. Interestingly, all ofthese areas have high annual rainfall, precipitation being predominant in the winter months (Mediterranean climates). In Hawaii, D. odorata occurs between 500 m and 2500 m above sea level in areas with 1250 mm to 2500 mm average annual rainfall. Invasion by D. odorata thus appears to have occurred only in locations with a high annual rainfall (and consequent moist conditions). Ideal growing conditions coupled with a lack ofnatural enemies may account for the invasive nature ofthis species in exotic regions. -133-

Chapter 9

References

Abbot, R. 1. (1986). Life history variation, polymorphism and outcrossing in Senecio vulgaris L.

Heredity 6,381-391.

Abbot, R. 1. & Forbes, D. G. (1993). Outcrossing rate and self-incompatibility in the colonizing species Senecio squalidus. Heredity 71, 155-159.

Abercrombie, M. , Hickrnan, M. , Johnson, M . L. & Thain, M. (1990). The New Penguin Dictionary ofBiology. Penguin Books Ltd., London.

Adair, R. 1. & Scott, 1. K. (1991). Distribution, life history, host specificity and suitability ofan undescribed Chrysolina species (Coleoptera) for biological control of Chrysanthemoides monilifera (Compositae). Bulletin ofEntomological Research 81, 235-242.

Adams, R. & Gianturco, M. (1956). Senecio alkaloids, the alkaloid from Senecio mikanioides. Journal ofthe American Chemical Society 79, 166-169.

Altmann,1. & Campbell, C. L. (1977). Effects ofherbicides on plant disease. AnnualReview of Phytopathology 15, 361-385.

Alvarez, M. E. (1997). Management ofCape Ivy (Delairea odorata) in the Golden Gate National Recreation Area. In Proceedings ofthe Exotic Pest Plant Council Symposium, vol. 3 (ed . M. Kelly, E . Wagner and P. Warner), pp. 91-95, California.

Alvarez, M . E. & Cushrnan, 1. H. (1997). Influence ofan invasive plant on species richness and composition ofthree coastal plant communities. Supplement to the Bulletin ofthe Ecological Society ofAmerica 78, 46-52. -134-

Amor, R. L. & Piggin, C. M . (1977). Factors influencing the establishment and success ofexotic plants in Australia. In Proceedings of the Ecological Society of Australia, vol. 10 (ed. D. Anderson), pp. 10-23. The Ecological Society ofAustralia, Melbourne.

Anderson, R. C., Loucks, O. L. & Swain, M. (1969). Herbaceous response to canopy cover, light intensity and through fall precipitation in conifer forests. Ecology 50, 256-263.

Andersson, S. (1988). Size dependent pollination efficiency in Anchusa officinalis

(Boraginaceae): causes and consequences. Oecologia 76, 125-130.

Andres, L. A, Davies, L. 1., Hams, P. & Wapshere, A 1. (1976). Biological control ofweeds. In Theory andpractice ofbiological control (ed. C. B. Huffaker and P. S. Messenger), pp. 481­ 500. Academic Press, New York.

Angevine, M . W. & Handel, S. N. (1986). Invasion offorest floor space, clonal architecture, and population growth in the perennial herb Clintonia borealis. Journal ofEcology 74,547-560.

Anon. (1971). Principles ofplant and pest control, vol 2. Subcommittee on weeds, Committee on Plant and Animal Pests, Agricultural Board, National Research Council. National Academy of Sciences.

Anon. (1996). Exotic pest plants ofgreatest ecological concern in California as ofAugust 1996. In CalEPPC News. California Exotic Pest Plant Council.

Archbald, G. (1995). Biology and control of German Ivy. California Department of Fish and Game Pesticide Seminar.

Archbald, G. (1997). Letter to David Chipping summarizing Cape Ivy fish toxicity experiments by Dr. Carla Bossard at St. Marys College. Conservation Projects, Golden Gate National Parks Association, San Francisco, California 94123 . -135-

Arnold, T. H. & de Wet, B. C. (1993). Plants ofsouthern Africa. National Botanical Institute,

South Africa.

Arroyo, M. T. K. (1976) . Geitonogamy in animal pollinated tropical angiosperms. A stimulus for the evolution ofself-incompatibility. Taxon 25, 543-548 .

Atwater, B. R. (1980). Germination, dormancy and morphology of the seeds of herbaceous ornamental plants. Seed Science and Technology 8, 523-573.

Baker, H. G. (1955). Self-compatibility and establishment after "long distance" dispersal. Evolution 9, 347-348.

Baker, H. G. (1965). Characteristics and modes oforigin ofweeds. In The genetics ofcolonising species (ed. H. G. Baker and G. L. Stebbins), pp. 588. Academic Press, New York.

Baker, H. G. (1974). The evolution ofweeds.AnnualReview ofEcology andSystematics 5, 1-24.

Balciunas,1. ( pers. comm .). Exotic and Invasive Weed Management Research Unit, USDA­ ARS Western Regional Research Centre, California.

Barraclough, D. (pers. comm.) . Natal Museum, 237 Loop Street, Pietermaritzburg, Natal.

Barret, P. & Richardson, M. (1986). Genetic attributes of invading species. In Ecology of biological invasions (ed. R. H. Groves and 1. 1. Burdon), pp 10-22. Cambridge University Press, London.

Bartlett, B. R. & van den Bosch, R. (1971). Foreign exploration for beneficial organisms. In

Biological control ofinsect pests andweeds (ed. C. B. Huffaker), pp. 283-345. Plenum Press, New York.

Bawa, K. S. (1974). Breeding system of tree species of a lowland tropical community. New Zealand Journal ofBotany 28, 85-92. -136-

Bazzaz, F. A. & Harper, 1. L. (1977). Demographic analysis ofgrowth ofLinum usitatissimum.

New Phytologist 78, 193-208.

Bell, A. D. (1984). Dynamic morphology: a contribution to plant population ecology. In Perspectives on plant population ecology (ed. R. Dirzo and 1. Sarukhan), pp. 48-65 . Sinauer

Associates, Massachusetts.

Bell, D. T. (1993). The effect of light quality on the germination of eight species from sandy habitats in Western Australia. Australian Journal ofBotany 41,321-326.

Bennet, M. B. (1987). Variation in genomic form in plants and its ecological implications. New Phytologist 106, 177-200.

Bennet, M. D. (1972). NuclearDNA content and minimum germination time.Proceedings ofthe Royal Society ofLondon 181, 109-135 .

Bennett, P. H. (1986). An investigation into the biological control of crofton weed, Ageratina adenophora (K.& R.) (syn. Eupatorium adenophorum Spreng.) by the gall flyProcecidochares uti/is Stone. PhD thesis, University ofNatal.

Bess, H. A. & Haramoto, F. H. (1959). Biological control of Pamakani, Eupatorium adenophorum, in Hawaii by a tephritid gall fly, Procecidochares utilis. 2. Population studies of the weed, the fly, and the parasites ofthe fly. Ecology 40, 244-249.

Beveridge, 1. L. & Wilise, C. P. (1959). Influence ofdepth ofplanting, seed size and variety on emergence and seedling vigour in alfalfa. Agronomy Journal SI, 731-734.

Bewley, 1. D. & Black, M. (1994). Seeds: Physiology ofdevelopment and germination, 2nd edition. Plenum Press, New York. -137-

Bicchi, c., Caniato, R., Tabacchi, R. & Tsoupras, G. (1989) . Capillary gas chromatography/ positive negative ion chemical ionisation mass spectrometry on pyrrolizidine alkaloids ofSenecio inaequidens using ammonia and hydroxyl ions as the reagent species.JournalofNaturalProducts 52,32-41.

Billet, E. E. and Burnett, J. H. (1978). The host parasite physiology ofthe maize smut fungus Ustilago maydis. Il, Translocation of 14C-Iabelled assimilates in smutted maize plants. Physiological Plant Pathology 12, 103-112.

Biller, A , Boppre, M., Witte, L. & Hartmann, T. (1994). Pyrrolizidine alkaloids in Chromalaena odorata: chemical and chemoecological aspects. Phytochemistry 35, 615-619.

Bjorkman, O. (1980) . Responses to different flux densities. In Encyclopedia ofplantphysiology, vol. 12A (ed . A Pirson and M. H. Zimmerman), pp. 57-107. Springer, Berlin.

Blossey, B., Schroeder, D., Right, S. D. & Maleck, R. A (1994). Host-specificity and environmental impact of the weevil Hylobius transversovittatus, a biological control agent of purple loosestrife (Lythrum salicaria). Weed Science 42, 128-133.

Borstel, K., Witte, L. & Hartmann, T. (1988). Pyrrolizidine alkaloid patterns in populations of Senecio vulgaris, S. vernalis and their hybrids. Phytochemistry 28, 1639-1643.

Bossard, C. & Benefield, C. (1995). War on German Ivy: Good news from the front. In Proceedings ofthe CaliforniaExoticPestPlant Council, Symposium (ed . J.E.Lovich, J.Randall and M. D. Kelly), pp. 63-64, California.

Brewbaker, J. L. & Majumder, S. K. (1961). Incompatibility and the pollen grain. Recent Advances in Botany 2, 1503-1508.

Brown, A H. D. & Marshall, D. R. (1980). The evolutionary genetics ofcolonising plants. In Proceedings ofthe SecondInternational Congress ofSystematics andEvolutionaryBiology (ed.

G. E. C. Scudder and J. L. Reveal), pp. 351-363. Hunt Institute for Botanical Documentation, Pittsburgh. -138-

Burdon, 1. 1. & Marshall, D. R. (1981). Biological control and the reproductive mode ofweeds. Journal ofAppliedEcology 18, 649-658 .

Carlquist, S. (1991). Anatomy ofvine and lianastems: a review and synthesis. In The biology of vines (ed. F. E. Putz and H.A. Mooney), pp 53-73 . CambridgeUniversityPress, New York.

Carpenter, F. L. (1976). Plant-pollinator interactions in Hawaii: pollination energetics in Metrosideros collina (Myrtaceae). Ecology 57, 1125-1144.

Carr, G.D. & Powell, E. A. (1986). Self-incompatibility inthe HawaiianMadiinae(Compositae): an exception to Baker's rule. Evolution 40, 430-434.

Catalano, S., Luschi, S., Flamini, G., Ciono, P. L., Neri, E. M. & Morelli, I. (1996). A xanthone from Senecio mikanioides leaves. Phytochemistry 42, 1605-1607.

CCWR (1999). Computing Centre for Water Research. University ofNatal, Private Bag X01, Scottsville, Pietermaritzburg.

Chalmers, A. H., Culvenor, C. C. 1. & Smith, L. W. (1965). Characteristics of pyrrolizidine alkaloids bygas, thin-layerand paper chromatography.Journal ojChromatography20,270-277.

Chipping, D. (1993). GermanIvy infestationin San Luis Obispo. In CalEPPC News, vol. 1, pp. 15-16 .

Christenson, L. D. & Foote, R. H. (1960). The biology of fruit flies . Annual Review of Entomology 5, 171-192.

Clement, S. L. & Cristofaro, M. (1995). Open-field tests in host-specificity determination of insects for biological control ofweeds. Biocontrol Science and Technology 5,395-406.

Cook, R. E. (1979). Patterns ofjuvenile mortality and recruitment in plants. In Topics in plant population biology (ed. O. T. lain, G. B. Johnson and P. H. Raven), pp. 207-231. Macmillan Press, London. -139-

Cook, R.E. (1985). Growth and development in clonal plant populations. In Population biology andevolution ofclonal organisms (ed . J. B. C. Jackson, L. W. Buss and R.E. Cook), pp. 59-78. Yale University Press, New Haven.

Copland, B . J. & Whelan, R. J. (1989). Seasonal variation in flowering intensity and pollination limitation offruit-set in four eo-occurring Banksia species. Journal ofEcology 77,509-523.

Cornell, H. V. & Hawkins, B. A.(1993). Accumulation ofnative parasitoid species on introduced herbivores: a comparison as hosts as natives and host as invaders. American Naturalist 141,847- 865.

Cousens, R. & Mortimer, M. (1995). Dynamics ofweed populations. Cambridge University Press, Cambridge.

Cox, D. (1998). Letter to Ramona Robison regarding Delairea in Australia. Department of Natural Resources and Environment. Victoria, Australia.

Crawley, M . J. (1989). The successes and failures ofweed biocontrol using insects. Biocontrol News andInformation 10, 213-223.

Cronk, Q. C. B. & Fuller, J. L. (1995). Plant invaders: the threat to natural ecosystems. Chapman and Hall, London.

Crowe, L. (1954). Incompatibility in Cosmos bipinnatus. Heredity 8, 1-11.

Cudney, D. W. & Hodel, D. (1986). German Ivy control in coastal sage scrub regions ofsouthern California. Report, Western Society ofWeed Science Research, San Diego.

Culvenor, C. C. J. & Geissman, T. A. (1961). The alkaloids of Senecio mikanioides QUo. Sarracine and Sarracine -oxide. Journal ofOrganic Chemistry 26, 3045-3050.

De Bach, P. (1964). Biological control ofinsect pests andweeds. Chapman and Hall, London. -140-

Daehler, C. C. (1998). The taxonomic distribution of invasive angiosperm plants: ecological insights and comparison to agricultural weeds. Biological Conservation 84, 167-180.

De Kroon, H. & Hutchings, M. 1. (1995). Morphological plasticity in clonal plants: the foraging concept reconsidered. Journal ofEcology 83, 143-152.

De Nettancourt, D. (1977). Incompatibility in Angiosperms. Springer-Verlag, Berlin.

De Steven, D. (1989). Genet and ramet demography ofOenocarpus mapora a clonal palm ofa Panamanian tropical moist forest. Journal ofEcology 77, 579-596.

Dean, S. 1., Holmes, P. M. & Weiss, P. W. (1986). Seed biology ofinvasive alien plants in South Africa and Namibia. In Proceedings ofthe National Symposium ofthe Ecology ofBiological Invasions (ed. L. A. McDonald, F. 1. Kruger and A. A. Ferrar), pp 157-178. Oxford University Press, Cape Town.

Dempster, 1. P. (1971). The population ecology of the cinnabar moth, Tyria jacobaeae L. (Lepidoptera: Arctiidae). Oecologia 7, 26-67.

Dennil, G.B. (1985). The effect ofthegall wasp Trichilogaster acacialongifoliae (Hymenoptera: Pteromalidae) on reproductive potential and vegetative growth ofthe weed Acacia longifolia. Agriculture, Ecosystems andEnvironment 14, 53-61.

Dennil, G. B. & Moran, V. C. (1989). On insect-plant associations in agriculture and the selection ofagents for weed biocontrol. Annals ofAppliedBiology 114, 157-166.

Dennil, G. B. & Hokkanen, H. M. T. (1990). Homeostasis and success in biological control of weeds - a question ofbalance. Agriculture, Ecosystems andEnvironment 33, 1-10.

Dodd, A. P. (1961). Biological control ofEupatorium adenophorum in Queensland. Australian Journal ofScience 23, 356-365. -141-

Dunn, P. H. & Campobasso, G. (1993) . Field test ofthe weevil Hadroplonthus trimaculatus and the fleabeetle Psylliodes chalcomera against musk thistle (Carduus nutans) . Weed Science 41, 656-663.

Edwards, P. B., Adair, R.1. & Holtkamp, R. H. (1996). Seedhead tephritids and the ~r parasitoids­ can predictions be made about likely parasitism of tephritids introduced into Australia for biological control of Asteraceae. In Proceedings of the IX International Symposium on Biological Control ofWeeds (ed. V. C. Moran and 1. H. Hoffmann), pp . 153-164. University of Cape Town, Stellenbosch, South Africa.

Elliot, W. (1994) . German Ivy engulfing riparian forests and heading for the uplands. In CalEPPC News, vol. 2, pp. 9.

Ellison, A. M . & Niklas, K. 1. (1988). Branching patterns of Salicornia europaea (Chenopodiaceae) at different successional stages: a comparison oftheoretical and real plants. American Journal ofBotany 75,501-512.

Elzen, G. W. (1983). Cytokinins and insect galls. Comparative Biochemistry andPhysiology 76, 17-19.

Engelbrecht, L. (1971). Cytokinin activity in larval infested leaves. Biochemie undPhysiologie der Plansen 162, 9-27.

Entwistle, P. F. (1983). Control ofinsects by virus diseases. Biocontrol News andInformation 4,203-220.

Eriksson, O. (1989). Seedling dynamics and life histories in clonal plants. Oikos 55,231-238.

Eriksson, O. (1992) . Evolution ofseed dispersal and recruitment in clonal plants. Oikos 63,439­ 448.

Eriksson, O. & Bremer, B. (1993). Genet dynamics ofthe clonal plant Rubus saxatilis. Journal ofEcology 81, 533-542 . -142-

Esau, K. (1960). Anatomy ofseedplants. John Wiley and Sons, New York.

Evans, 1. P . & Cain, M. L. (1995). A spatially explicit test offoraging behaviour in clonal plants.

Ecology 76, 1147-1155.

Evetts, L. L. & Burnside, G. C. (1972). Germination and seedling development of common milkweed and other species. Weed Science 20, 371-374.

Fagg, P. C. (1989). Control ofDelairea odorata (Cape Ivy) in native forest with the herbicide clopyralid. Plant Protection Quarterly 4, 107-110.

FAO (1997). Code of conduct for the import and release of exotic biological control agents . Biocontrol News andInformation 18, 119-124.

Faucroy, M. and Braun, C. (1967). Observations sur la galle de l' Aulax glenchamae L. sur Glenchoma hederacea L. - seconde partie-histologie et role physiologique de la coque sclrifie. Marcellia 34, 3-30.

Fisher, A. (1997). The effects of Senecio mikanioides on riparian invertebrate populations in Redwood Creek and Elk Creek in the Golden GateNational Recreation Area. Unpublishedpaper.

Fitter, A. H. & Hay, R. K. M. (1981). Environmental physiology ofplants. Academic Press, London.

Forbert, M. (1998). Some tips on herbicide treatment of Cape Ivy (Delairea odorata). In CalEPPC News, vol. 6, pp. 4.

Forsyth, C. & Brown, N. A. C. (1982). GerminationofBidenspilosaL. The NewPhytologist 90, 150-164

Freeman, T. E. & Charudattan, R. (1984). Conflicts in the use ofplant pathogens as biocontrol agents for weeds. In Proceedings ofthe VI International Symposium on the Biological Control ofWeeds (ed. E. S. Delfosse), pp. 351-357. Agriculture Canada, Vancouver. -143-

Frei, H., Luthy, 1., Brauchli, 1., Zweifel, u., Wurgler, F. E. & Schlatter, C. (1992). Structure/activity relationships of the genotoxic potencies of sixteen pyrrolizidine alkaloids assayed for induction of somatic mutation and recombination in wing cells of Drosophila melanogaster. Chemical andBiological Interactions 83, 1-22.

Freidberg, A. (1984). Gall Tephritidae (Diptera). In The biology of gall insects (ed. 1. N . Ananthakrishnan), pp . 129-167. Edward Arnold, New Delhi.

Fritz, R. S. & Simms, E. L. (1992). Plant resistance to herbivores andpathogens - ecology, evolution andgenetics. University ofChicago Press, Chicago.

Gillet, 1. D., Barley, K. L. S., Kassulke, R. C. & Miranda, H. 1. (1991). Natural enemies ofSida acuta and S. rhombifolia (Malvaceae) in Mexico and their potential for biological control ofthese weeds in Australia. Environmental Entomology 20, 882-888.

Goeden, R. D. (1988). A cap sule history ofbiological control ofweeds. Biocontrol News and Information 9,55-61.

Goldingay, R. L. & Whelan, R. 1. (1990). Breeding system and tests for pollen-limitation in two species ofBanksia. Australian Journal ofBotany 38,63-71.

Grime, 1. P. (1966). Shade avoidance and shade tolerance in flowering plants. British Ecological Society Symposium 6, 187-207.

Grime, 1. P., Mason, G., Curtis, A. V., Rodman, 1., Band, S. R., Mowforth, M. A. G., Neal, A. M. & Shaw, S. (1981). A comparative study of germination characteristics in a local flora. Journal ofEcology 69, 1017-1059.

Grobbelaar, E., Neser, S. & Neser, O. C. (1999). Biological control ofDelairea odorata in California: a survey for its natural enemies in South Africa. In Proceedings of the Twelfth

Entomological Congress (ed. 1. B. 1. van Rensburg and 1. van den Berg), pp. 52. Entomological Society of Southern Africa, Potchefstroom. -144-

Gross, R. S. & Werner, P. A. (1983). Relationships among flowering phenology, insect visitors, and seed set of individuals: experimental studies on four eo-occurring species of goldenrod

(Solidago: Compositae). Ecological Monographs 53, 95-117.

Groves, R. H . (1986). Plant invasions of Australia - an overview. In Ecology of biological invasions (ed . R. H. Groves and J. J. Byrdon), pp. 155-163. Cambridge University Press, New

York.

Handel, S. N . (1985). The intrusion of clonal growth patterns on plant breeding systems. American Naturalist 125, 367-384 .

Harper, J. L. (1977). Population biology ofplants. Academic Press, London.

Harper, J. L., Lovell, P. H. & Moore, K. G. (1970). The shapes and sizes ofseeds. Annual Review ofEcology and Systematics 1, 327-356.

Harris, P. (1973) . The selection ofeffective agents for the biological control ofweeds. Canadian Entomologist 105, 1495-1503.

Harris, P. (1989). The use ofTephritidae for the biological control ofweeds. Biocontrol News and Information 10, 7-16.

Harris, P. (1991). Classical biocontrol ofweeds: its definition, selection ofeffective agents, and administrative-political problems. Canadian Entomologist 123, 827-849.

Harris, P . & Zwolfer, H. (1968). Screening of phytophagous insects for biological control of weeds. Canadian Entomologist 100, 295-303.

Hartmann, T. (1996). Diversity and variability ofplant secondary metabolism: a mechanistic view. Entomologia Experimentalis et Applicata 80, 177-188.

Hartmann, T. (1999). Chemical ecology ofpyrrolizidine alkaloids. Planta 207,483-495. -145-

Hartmann, T. & Toppel, G. (1987). Senecionine -oxide the primary product of pyrrolizidine alkaloid biosynthesis in root cultures ofSenecio vulgaris. Phytochemistry 26, 1639-1643.

Hartmann, T. & Zimmer, M. (1986) Organ-specific distribution and accumulation ofpyrrolizidine alkaloids during the life history oftwo annual Senecio species. Journal ofPlantPhysiology 122, 67-80.

Hartmann, T., Witte, L., Ehmke, A, Theuring, c., Rowell-Rahier, M. & Pasteels, 1. M. (1997). Selective sequestration and metabolism ofplant derived pyrrolizidine alkaloids by chrysomelid leaf beetles. Phytochemistry 45,498-497.

HEAR (Hawaiian Ecosystems at Risk Project") (1986). Alien species map index. University of Hawaii, Botany Department, Hawaii.

Hendricks, 1. D., Sinnhuber, R. 0., Henderson, M. C. & Buhler, D. R. (1981). Liver and kidney pathology in rainbow trout (Salmo gairdneri) exposed to dietary pyrrolizidine (Senecio) alkaloids. Experimental andMolecular Pathology 35, 170-183 .

Hessing, M. B. (1988). Geitonogamous pollination and its consequences In Geranium caespitosum. American Journal ofBotany 75, 1324-1333.

Heywood, V. H. (1989). Patterns, extents and modes of invasions by terrestrial plants. In Biological invasions: a globalperspective (ed. 1. A Drake, H. A Mooney, F. Di Castri, R. H. Groves, F. 1. Kruger, M. Rejmanek and M. Williamson), pp. 31-43. John Wiley, Chichester.

Heywood, V. H. (1993). Flowering plants ofthe world. Oxford University Press, New York.

Heywood, V. H., Harbourne, J. B. & Turner, B. L. (1977). The biology and chemistry ofthe Compositae, vol. 1. Academic Press, New York.

Billiard, O. M . (1977). Compositae ofNatal. University ofNatal Press, Pietermaritzburg. -146-

Hoffmeister, T. (1992). Factors determining the structure and diversity ofparasito id complexes in tephritid fiuit flies. Oecologia 89, 288-297.

Holm, L. G., Doll, 1., Holm, E ., Pancho, 1. & Herberger, 1. (1997). Worldweeds: naturalhistory anddistribution. John Wiley and Sons Inc ., London.

Holm, L. G., Pancho, 1. v., Herberger, 1. P. & Plucknett, D . L. (1979). A geographical atlas of worldweeds. Wi1ey Interscience, New York.

Horowitz, M. & Taylorson, R. B. (1982). Effect ofhigh temperatures on imbibition, germination, and thermal death ofvelvetleaf(Abutilon theiphrasti).Canadian Journal ofBotany 61,2269­ 2276.

Houssard, G. & Escarre, 1. (1995). Variation and covariation among life history traits ofRumex acetosella from a successional old-field gradient. Oecologia 102, 70-80.

Huffaker, C. B . (1964). Fundamentals ofbiological weed control. In Biologicalcontrolofinsects pests andweeds (ed. P. DeBach), pp. 631-663. Chapman and Hall, London.

Huffaker, C. B. , Messenger, P. S. & DeBach, P. (1971). The natural enemy component. In Biological Control (ed . C. B. Huffaker), pp. 16-68 . Plenum Press, New York.

Hutchings, M. J. (1988). Differential foraging for resources and structural plasticity in plants. Trends in Ecology andEvolution 3, 200-204.

Hutchings,M. 1. & De Kroon, H. (1994). Foraging in plants: the role ofmorphological plasticity in resource acquisition. Advances in Ecological Research 25, 159-238.

Jacobi, 1. D. & Warshauer, F. R. (1992). Distribution ofsix alien plant species in upland habits on the island ofHawaii. In Alien plant invasions in native ecosystems ofHawaii: Management andresearch (ed . C. P. Stone, C. W. Smith and 1. T. Tunison), pp. 155-180. University ofHawaii Press, Honolulu. -147-

James, R. R., McEvoy, P. B. & Cox, C. S. (1992). Combining the cinnabar moth (Tyria jacobaeae) and the ragwort fleabeetle (Longitarsus jacobaeae) for control ofragwort (Senecio jacobaeae): an experimental analysis. Journal ofAppliedEcology 29,589-596.

Jeffrey, C. (1986). The Senecioneae in east tropical Africa. Kew Bulletin 41, 873-937.

Jeffrey, C. & Chen, Y. L. (1984). Taxonomic studies in the tribe Senecioneae (Compositae) of Eastern Asia. Kew Bulletin 39, 264-446.

Johnson, A. E. , Molyneux, R. 1. & Merrill, G. B. (1985). Chemistry of toxic range plants. Variation in pyrrolizidine alkaloid content ofSenecio, Amsinckia, and Crotolaria species.Journal ofAgriculture andFood Chemistry 33,50-55.

Julien, M. H. (1989). Biological control ofweeds worldwide: trends, rates of success and the future . Biocontrol News andInformation 10,299-306.

Julien, M. H. (1992). Biological control ofweeds: a world catalogue ofagents andtheir target weeds, 3rd edition. CAB International, Wallingford.

Kays, S. & Harper, 1. L. (1974). The regulation ofplant and tiller density. Journal ofEcology 62, 97-105 .

Keeley, P. E. & Thullen, R. 1. (1978). Light requirement ofyellow nutsedge (Cyperus esculentus) and light interception by crops. Weed Science 26, 10-16.

King, L. 1. (1966). Weeds ofthe world: biology and control. Interscience Publishers Inc., New York.

Klingman, K. C., Ashton, F. M. & Noordhoff, L. 1. (1982). Weed science: principles and practices, pp. 450. John Wiley and Sons, New York. -148-

Koekemoer, M. J. & Warren, F. L. (1951). The Senecio alkaloids. Part VIII. The occurrence and preparation ofthe N-oxides. Animproved method ofextraction oftheSenecio alkaloids. Journal ofthe Chemical Society 5,66-68.

Koller, D . & Roth, N . (1964). Studies on the ecological and physiological significance of amphicarpy in Gymnarrhena micrantha (Compositae). American Journal ofBotany 51, 26-35 .

Krebs, H. c., Carl, T. & Habermehl, G. (1996). Pyrrolizidine alkaloid composition in six Brazilian Senecio species. Phytochemistry 43, 1227-1229.

Kruger, F. J. (1977). Invasive woody plants in the Cape fynbos with special reference to the biology and control ofPinus pinaster. In Proceedings 2ndNational Weed Conference ofSouth Africa (ed. A. A. Balkema), pp. 57-74. Cape Town, South Africa.

Lalonde, R. G. and Shorthouse, J.D. (1984). Developmental morphology ofthegall ofUrophora cardui (Diptera, Tephritidae) in the stems ofCanadathistle (Cirsium arvense) . Canadian Journal ofBotany 62, 1372-1384.

Landolt, P. J. & Phillips, T. W. (1997). Host plant influences on sex pheromone behaviour of phytophagous insects. AnnualReview ofEntomology 42, 371-391.

Lauer, E. (1953). Uber die Keimtemperatuur van Ackerunk-rautem und derem einfluss aufdie Eusammetisetzung van Unkrautgesellschaften. Flora oder Algemeine. Botanische Zeitung 140, 551-595.

Lawrence, M. E. (1980). Senecio L. (Asteraceae) in Australia: chromosome numbers and the occurrence ofpolyploidy. Australian Journal ofBotany 28, 151-165.

Lawrence, M . E. (1985). Senecio L. (Asteraceae) in Australia: reproductive biology ofa genus found primarily in unstable environments. Australian Journal ofBotany 33, 197-208. -149-

Lawton, 1. H. (1984). Ecological theory and choice ofbiological control agents. In Proceedings ofthe VI International Symposium on the Biological Control ofWeeds (ed. E. s.Delfosse), pp. 13-26. Agriculture Canada, Vancouver.

Leakey, R. R. B. (1981). Adaptive biology of vegetatively regenerating weeds. Advances in AppliedBiology 6,57-90.

Levin, D . A. & Kerster, H. W. (1974). Gene flow in seed plants. Evolutionary Biology 7, 139­ 220.

Lovett Doust, L. (1981). Intraclonal variation and competition in Ranunculus repens. New Phytologist 89, 495-502.

Lovett Doust, L. (1987). Population dynamics and local specialisation in a clonal perennial

(Ranunculus repens). Ill. Responses to light and nutrient supply. Journal ofEcology 75,555­ 568.

Luthy, 1., Zweifel, u., Karlhuber, B. & Schlatter, C. (1981). Pyrrolizidine alkaloids ofSenecio alpinus L. and their detection in feedstuffs. Journal ofAgriculture andFoodChemistry 29,302­ 305.

Mack, R. N. (1996). Predicting the identity and fate ofplant invaders: emergent and emerging approaches. Biological Conservation 78, 107-121.

Maddox, D. M. & Sobhain, R. (1987). Field experiment to determine host-specificity and oviposition behaviour of Bangasternus orientalis and Bangasternus fausti (Coleoptera: Curculionidae), biological control candidates for yellow starthistle and diffuse knapweed. Environmental Entomology 16, 645-648.

Mansell, M . (pers. comm.). Plant ProtectionResearchInstitute, PrivateBag X134, Pretoria 0001, South Africa. -150-

Marohasy. J. 1. (1989). A survey offireweed (Senecio madagascariensis Poir) and its natural enemies in Madagascar with a view to biological control in Australia. PlantProtection Quarterly

4, 139-140.

Mattocks, A. R. (1967). Spectrophotometric determination ofunsaturated pyrrolizidine alkaloids.

Analytical Chemistry 39, 443-447.

Mattocks, A. R. (1986). Chemistry and toxicology ofpyrrolizidine alkaloids. Academic Press

Inc., London.

Mayer, A. M. & Poljakoff-Mayber, A. (1989) . The germination ofseeds, 4th edition. Pergamon

Press, New York.

McClay, A. S. (1989). Selection ofsuitable target weeds for classical biocontrol in Alberta. PhD thesis, Alberta Environmental Centre.

McClay, A. S. & Palmer, W. A. (1995). Phytophagous arthropods associated withParthenium hystophorus (Asteraceae) in North America. Environmental Entomology 24, 796-810.

McEnvoy, P. B. (1984). Depression in ragwort (Senecio jacobaeae) abundance following introduction of Tyria jacobaea and Longitarsus jacobaeae on the central coast of Oregon. In Proceedings ofthe VI International Symposium on the Biological Control ofWeeds (ed. E. s. Delfosse), pp. 57-64. Agriculture Canada, Vancouver.

McEnvoy, P. B., Rudd, N. T., Cox, C. S. & Huso, M. (1993). Disturbance, competition, and herbivory effects on ragwort Senecio jacobaeae populations. EcologicalMonographs 63, 55-75.

MeVean, D. (1966). Ecology of Chondrilla juncea L. in South Eastern Australia. Journal of Ecology 54,345-365.

Minitab? (1995). Minitab for Windows. Minitab Inc., 3081 Enterprise Drive, State College, PA 16801-3008. -151-

Mitchell, P. L. & Woodward, F. L. (1988). Responses of three woodland herbs to reduced photosynthetically active radiation and low red to far-red ratio in shade. Journal ofEcology 76,

807-825.

Molyneux, R. 1., Johnson, A. E ., Roitman, 1. N . & Benson, M. E. (1979). Chemistry oftoxic range plants. Determination ofpyrrolizidine alkaloid content and composition in Senecio species by nuclear magnetic resonance spectroscopy. Journal ofAgriculture and Food Chemistry 27,

494-499.

Molyneux, R. J. & Ralphs, M . H. (1992). Plant toxins and palatability to herbivores. Journal of

Range Management 45, 13-1 8.

Montenego, G., Teillier, S., Arce, P. & Poblete, V. (1991). Introduction ofplants into Chile. In Biogeography ofmediterranean invasions (ed . R. H. Groves and F. D. Di Castri), pp. 115-131 .

Cambridge University Press, New York.

Mooney, H. A. & Chiariello, N . (1984). The study of plant function: the plant as a balanced system. In Perspectives on plantpopulation ecology (ed. R. Dirzo and 1. Sarukhan), pp. 305-323.

Sinauer, Massaschussets.

Moore, K. (1997). Battling the Kudzu ofthe west: controlling Cape Ivy (formerly German Ivy) by hand removal. In CalEPPC News, vol. 5, pp. 4.

Morin, L., Hill, R. L. & Matayoshi, S. (1997). Hawaii's successful biological control strategy for mist flower (Ageratina riparia) - can it be transferred to New Zealand? Biocontrol News and Information 18, 77-87.

Mortimer, A. M. (1990). The biology ofweeds. In Weed control handbook: principles (ed. R. 1. Hance and K. Holly), pp. 1-42. Blackwell Scientific Publications, Oxford.

Mott,1. 1. (1972). Germination studies on some annual species from an arid region ofWestem Australia. Journal ofEcology 60, 293-304. -152-

Muller, H (1989). Structural analysis ofphytophagous insect guilds associated with the roots of Centaurea maculosa Lam., C. diffusa Lam and C. vallesiaca Jordon in Europe. Oecologia 78,

41-52.

Muller-Scharer, H , Lewinsohn, T. M. & Lawton, J. H (1991). Searching for weed biocontrol agents - when to move on? Biocontrol Science and Technology 1,271-280.

Munro, H K. (1940). Further South African gall forming Trypetidae (Diptera) with description ofnew species. Journal ofthe Entomological Society ofSouth Africa 3, 76-87.

Munro, H K. (1952). A remarkable new gall forming trypetid (Diptera) from southern Africa and its allies. South African Department ofAgriculture Entomological Memoirs 2, 329-341 .

Munro, H. K. (1953). Records of some Trypetidae (Diptera) collected on the Bernard Carp Expedition to Barotseland, 1952, with a new species from Kenya. Journal ofthe Entomological Society ofSouth Africa 16, 217-226.

Munro, H K. (1957). Trypetidae. Ruwenzori Expedition 9,853-1054.

Noble, I. R. (1989). Attributes ofinvaders and the invading process. In Biological Invasions: a globalperspective (ed. J. A. Drake, H. A. Mooney, F. Di Castri, R. H Groves, F. J. Kruger, M. Rejmanek and M. Williamson), pp 301-313. John Wiley, Chichester.

O'Brien, C. W. (1995). Curculionidae as premier biocontrol agents (Coleoptera: Curculionidae). Memoirs ofthe Entomological Society ofWashington 14, 119-128.

Panetta, F. D. (1977). The effects ofshade on seedling growth in the groundsel bush (Baccharis halimifolia L.). Australian Journal ofAgricultural Research 28,681-690.

Panetta, F. D. & Mitchell, N. D. (1991). Homocline analysis and the prediction ofweediness. Weed Research 31, 273-284. -153-

Perrins, 1., Williamson, M. & Fitter, A.(1992). A survey ofdiffering views ofweed classification: implications for regulation ofintroduction. Biological Conservation 60, 47-56.

Pitkella, L. F., Stanton, D. S. & Peckenham, M. O. (1980). Effects of light and density on resource allocation in a forest herb, Asteracuminatus (Compositae). AmericanJournal ofBotany

67,942-948.

Plummer,1. A. & Bell, D. T. (1995). The effect oftemperature, light and gibberellic acid (GA3) on the germination ofAustralian everlasting daises (Asteraceae, tribe Inuleae). AustralianJournal ofBotany 43,93-102.

Pons, T. L. (1992). Seed responses to light. In The ecology ofregeneration in plant communities (ed. M. Fenner), pp. 259-284. CAB International, Wallingford.

Pooley, E. (1993). The completejieldguide to trees ofNatal, l" edition . Natal Flora Publication Trust, Durban.

Popay, A. I. & Roberts, E. H. (1970). Factors involved in the dormancy and germination of Capsella bursa-pastoris (L.) Medik. and Senecio vulgaris L. Journal ofEcology 58, 103-122.

Prach, K. & Wade, P. M. (1992). Population characteristics ofexpansive perennial herbs.Preslia 64,45-51.

Price, S. C. & lain, S. K. (1981). Are inbreeders better colonisers? Oecologia 49, 283-286.

Pysek, P. (1995). On the terminology used in plant invasion studies. In Plant invasions-general aspects and specialproblems (ed. P. Pysek, K. Prach, M. Rejmanek and M. Wade), pp. 71-81. SPB Academic Publishing, Amsterdam.

Pysek, P., Prach, K. & Smilauer, P. (1995). Relating invasion success to plant traits: an analysis ofthe Czech alien flora . In Plant invasions - general aspects andspecialproblems (ed. P. Pysek, K. Prach, M. Rejmanek and M. Wade), pp. 30-43 . SPB Academic Publishing, Amsterdam. -154-

Rathcke, B. (1983). Competition and facilitation among plants for pollination. In Pollination Biology (ed. L. Reel), pp. 305-329. Academic Press, New York.

Rejmanek, M. (1989). Invasibility of plant communities. In Biological invasions. A global perspective (ed. 1. A. Drake,H. A. Mooney, F. Di Castri, F. Groves, R. H. Kruger, M. Rejmanek and M. Williamson), pp. 369-388. John Wiley, Chichester.

Rejmanek, M. (1995). What makes a species invasive? In Plant invasions-general aspects and specialproblems(ed. P. Pysek, K. Prach, M. Rejmanek andM. Wade), pp. 3-13. SPB Academic Publishing, Amsterdam.

Rejmanek, M. (1996). A theory of seed plant invasiveness: the first sketch. Biological Conservation 78,171-181.

Rejmanek, M. & Richardson, D. M. (1996). What attributes make some plant species more invasive? Ecology 77, 1655-1659.

Rejmanek, M., Thomsen, C. D. & Peters, I. D. (1991). Invasive vascular plants ofCalifornia. In Biogeography ofmediterranean invasions (ed. R. H. Groves and F. Di Castri), pp. 81-101. Cambridge UniversityPress, New York.

Richards, A. 1. (1986). Plant breeding systems. AlIen & Unwin, London.

Robins, D. 1. (1982). The pyrrolizidine alkaloids. Natural Product Reports 41, 115-203.

Robison, M. (1999). Proposal for the control ofDelairea odorata (Cape Ivy) in California. PhD thesis proposal, UniversityofCalifornia, Davis.

Robison, M.(pers. comm.). University ofCalifornia, Davis, CA 95616, California, USA.

Robocker, W. C. (1977). Germination ofseedsofcommonyarrow (Achillea millefolium) and its herbicidal control. Journal ofRange Management 25,456-458. -155-

Roeder, E. (1990). Carbon-13 NMR spectroscopy ofpyrrolizidine alkaloids. Phytochemistry 29,

11-29.

Roggen, H. P . & van Dijk, A.J. (1972). Breaking incompatibility in Brassica oleracea L. by steel

brush pollination. Euphytica 21, 48-51 .

Rokich, D . P. & Bell, D. T. (1995). Light quality intensity effects on the germination ofspecies from the Jarrah (Eucalyptus marginata) forest of Western Australia. Australian Journal of

Botany 43, 169-179..

Romeo, J. T. , Saunders, J. A. & Barbosa, P. (1995). Phytochemical diversity and redundancy in

ecological interactions. Recent Advances in Phytochemistry 30, 1-319.

Rothschild, M . (1973). Secondary plant substances and warning colouration in insects. In Insect/plant relationships (ed . H. F. van Emden), pp. 59-83 . Blackwell, Oxford.

Salonen, V. (1994). Growth and reproduction ofRubus saxatilis L. in relation to availability of

light. Acta Ecologica 15, 484-493 .

Schroeder, D . (1985). The search for effective biological control agents in Europe. 1. Diffuse and spotted knapweed. In Proceedings ofthe 6th International Symposium on the Biological Control ofWeeds (ed . E. S. Delfosse), pp. 103-119. Agriculture Canada, Vancouver.

Schroeder, D . & Goeden, R. D. (1986). The search for natural enemies ofintroduced weeds for biological control - in theory and practice. Biocontrol News andInformation 7, 147­ 155.

Scott, J. K. & Way, M . K. (1990). A survey in South Africa for potential biological control agents against capeweed, Arctotheca calendula (L.) Levyns (Asteraceae). Plant Protection Quarterly 5,31-33.

Scott, J. K. & Delfosse, E. S. (1992). Southern African plants naturalised in Australia: a review ofweed status and biological control potential. Plant Protection Quarterly 7, 70-80. -156-

Sheldon, J. C. (1974). The behaviour ofseeds in soil. Ill. The influence ofseed morphology and the behaviour of seedlings on the establishment ofplants from surface lying seeds. Journal of

Ecology 62, 47-66.

Shepherd, R. C. H. (1993). The role ofbiological control in weed management. PlantProtection

Quarterly 8, 144-146.

Slade, A J. & Hutchings, M . J. (1987a). The effects of nutrient availability on foraging in the clonal herb Glechoma hederacea. Journal ofEcology 75,95-112.

Slade, A J. & Hutchings, M. J. (1987b). The effects oflight intensity on foraging in the clonal herb Glechoma hederacea. Journal ofEcology 75,639-650.

Slade, A J. & Hutchings, M. J. (1987c). Clonal growth and plasticity in foraging behaviour in Glechoma hederacea. Journal ofEcology 75, 1023-1036.

Smreciu, E. A, Currah, R. S. & Toop, E . (1988). Viability and germination of herbaceous perennial species native to the southern Alberta grasslands. Canadian FieldNaturalist 102, 31­ 38.

Sobrevila, C. (1989). Effects of pollen donors on seed formation in Espletia schultzii (Compositae) populations at different altitudes. Plant Systematics andEvolution 166,45-67.

Solangaarachchi, S. M. & Harper, J. L. (1987). The effect ofcanopy filtered light on the growth ofthe white clover Trifolium repens. Oecologia (Berl.) 72,372-376.

Stelljes, M. E. & Seiber, J. N. (1990). Pyrrolizidine alkaloids in an overwintering population of monarch butterflies (Danaus plexippus) in California. Journal ojChemical Ecology 16, 1459­ 1470.

Steel, R.G.D. & Torrie, J.H. (1980). Principles and procedures oj statistics: a biometrical approach. McGraw-Hill International Editions, Singapore. -157-

Stelljes, M. E., Kelley, R. B., Molyneux, R. J. & Seiber, J. N. (1991). GC-MS determination of pyrrolizidine alkaloids in four Senecio species . Journal ofNatural Products 54, 759-773.

Stephens, R. J. (1982). Theory andpractice ofweedcontrol. The Macmillan Press Ltd, London.

Sultan, S. E. (1987). Evolutionary implications ofphenotypic plasticity in plants. Evolutionary

Biology 21, 127-176.

Sultan, S. E. & Bazzaz, F. A. (1993). Phenotypic plasticity inPolygonumpersicaria. 1.Diversity and uniformity in genotypic norms ofreaction to light. Evolution 47, 1009-1031.

Surles, W. & Kok, T. T. (1978). Carduus thistle seed destruction by Rhinocyllus conicus. Weed Science 26, 264-269.

Sutherland, W. J. & Stillman, R. A. (1988). The foraging tactics ofplants. Oikos 52,239-244.

Syrett, P. (1985). Host-specificity oftheragwort flea beetleLongitarsusjacobaeae (Waterhouse) (Coleoptera: Chrysomelidae). New Zealand Journal ofZoology 12, 335-340.

Syrett, P., Fowler, S. V. & Emberson, R. M. (1996). Are chrysomelid beetles effective agents for biological control ofweeds? In Proceedings ofthe IX international symposium on biological control ofweeds (ed. V. C. Moran and J. H. Hoffmann), pp. 399-407. University ofCape Town, Stellenbosch, South Africa.

Takahashi, N. (1979). A concept ofdormancy, germination and longevity in cereal seeds. Science Reports Research Institutes Tohoku University 30,21-27.

Thompson, J. N. & Willson, M. F. (1978). Disturbance and the dispersal offleshy fruits. Science 200, 1161-1163.

Thompson, K. (1990) . Genome size, seed size and germination temperature in herbaceous angiosperms. Evolutionary Trends in Plants 4, 113-116. -158-

Timbilla J. A & Braimah, H. (1996). The menace ofChromolaena odorata: a case study in the Brong Ahafo region of Ghana. In Proceedings of the Third International Workshop on the Biological Control andManagement ofChromolaena odorata (ed. U. K. Prasad, P. Muniappan,

P. Ferrar and 1. P. Aeschlimann), pp. 67-75. Orstam, Abijan, Cote d'Ivoire.

Toppel, G., Witte, L., Riebesehl, B., Borstel, K. & Hartmann, T. (1987). Alkaloid patterns and biosynthetic capacity ofroot culturesfrom some pyrrolizidine alkaloid producingSenecio species.

Plant Cell Reports 6, 466-469.

Trujilo, E. E. (1984). Biological control ofhamakua pamakani with Cercosporella sp. in Hawaii. In Proceedings ofthe VI International Symposium on the Biological Control ofWeeds (ed. E. S. Delfosse), pp. 661-671. Agriculture Canada, Vancouver.

USDA (United States Department OfAgriculture) (1998). Germination enhancement trials with Cape Ivy seeds. Agricultural Research Service Invasive Weed Seed Laboratory, Reno. van Dam, N. M. & Vrieling, K. (1994). Genetic variation in constitutive and inducible pyrrolizidine alkaloid levels in Cynoglossum officinale L. Oecologia 99, 374-378. van der Meijden, E. (1996). Plant defence, an evolutionary dilemma: contrasting effects of (specialist and generalist) herbivores and natural enemies. Entomologia Experimentalis et Applicata 80, 307-310. van Driesche, R. G. & Bellows, T. S. (1996). Biological control. Chapman and Hall, New York. van Staden, 1., Davey, 1. E. and Noel, AR A (1977). Gall formation in Erythrina latissima. Zeitschriftfur Planzenphysiologie 84, 283-294.

Vaughton, G. (1988). Pollination and seed set ofBanksia spinulosa: evidence for autogamy. Australian Journal ofBotany 36,633-642.

Vermeij, G. (1996). An agenda for invasion biology. Biological Conservation 78, 3-9. -159-

Vogt Anderson, U. (1995). Comparison ofdispersal strategies ofalien and native species in the Danish flora. In Plant Invasions - general aspects andspecialproblems (ed. P. Pysek, K. Prach, M. Rejmanek and W. Wade), pp. 45-57. SPB Academic Publishing, Amsterdam.

Vrieling, K. & van Vyk, A. M. (1994). Estimating costs and benefits ofthe pyrrolizidine alkaloids ofSenecio jacobaeae under natural conditions. Oikos 70,449-454.

Wan, F., Harris, P., Cai, L. & Zhang, M. (1996). Host-specificity ofAltica carduorum Guer. (Chrysomelidae: Coleoptera), a defoliator of Cirsium arvense (L.) Scop . (Asteraceae) from north-western China. Biocontrol Science and Technology 6,521-530.

Wapshere, A. J. (1974). A strategy for evaluating the safety of organisms for biological weed control. Annals ofAppliedBiology 77, 201-211 .

Wapshere, A. 1. (1989). A testing sequence for reducing rejection ofpotential biological control agents for weeds. Annals ofAppliedBiology 114, 515-526.

Waser, N. M . & Price, M. V. (1989). Optimal outcrossing in Ipomopsis aggregata: seed set and offspring fitness . Evolution 43, 1097-1109.

Watson, A. K. (1984). Host-specificity of plant pathogens in biological weed control. In Proceedings ofthe VI International Symposium on the Biological Control ofWeeds (ed. E. S. Delfosse), pp. 577-586. Agriculture Canada, Vancouver.

Wells, M. 1. & Stirton, C. 1. (1982). South Afiica. In Biology and ecology ofweeds (ed. W. Holzner and M. Numata), pp. 339-345. Junk, The Hague.

Werner, P. A. (1975). Predictions of fate from rosette size in teasel (Dipsacus fullonum). Oecologia 20, 197-201.

Whelan, R. J. & Goldingay, R. L. (1989). Factors affecting fruit-set in Telepea speciosissima (proteaceae): the importance ofpollen-limitation. Journal ofEcology 77, 1123-1134. -160-

Whelan, R. J. & Goldingay, R. L. (1986). Do pollinators influence seed-set in Banksiapaludosa Srn. and Banksia spinulosa R. Br.? Australian Journal ofEcology 77, 1123-1134.

Whigham, D. (1974). An ecological life history study of Uvularia perfoliata L. American

MidlandNaturalist 91, 343-359.

Widen, B. & Widen, M. (1990). Pollen limitation and distance-dependent fecundity in females in the gynodioecious herb Glechoma hederacea (Lamiaceae). Oecologia 83, 191-196.

Williamson, M. H. & Fitter, A. (1996). The characters of successful invaders. Biological

Conservation 78, 163-170.

Willis, A. J. & Groves, R. H. (1991). Temperature and light effects on the germination ofseven native forbes. Australian Journal ofBotany 39, 219-228.

Witte, L., Rubiolo, P., Bicchi, C. & Hartmann, T. (1993). Comparative analysis ofpyrrolizidine alkaloids from natural sources by gas chromatography-mass spectrometry. Phytochemistry 32, 187-196.

Woodburn, T. L. (1993). Host-specificity testing, release and establishment of Urophora solstitialis L. (Diptera: Tephritidae), a potential biological control agent for Carduus nutans L., in Australia. Biocontrol Science and Technology 3,419-426.

Yano, K. & Heppner, J. (1983). Description of hamakua pamakani plume moth from Hawaii (Lepidoptera: Pterophoridae). Proceedings ofthe Hawaiian EntomologicalSociety 24, 335-339.

Zapata, T. R. & Arroyo, M. T. K. (1978). Plant reproductive ecology of secondary deciduous tropical forests in Venezuela. Biotropica 10, 221-230.

Zar, J.H. (1974). Biostatistical analysis. Prentice Hall Inc., London.

Zimdahl, R. L. (1993). Fundamentals ofweed science. Academic Press Inc., New York. -161-

Zimmerman, M. & Pyke, G. H. (1988). Reproduction in Polemonium: assessing the factors limiting seed set. American Naturalist 131, 723-738.

Zwolfer, H. (1974). Das Treffpunkt-Prinzip als Kommunikationsstrategie and Isolations mechanismus bei Bohrfliegen (Diptera: Trypeidae). Entomologica Germanica 1, 11-20.

Zwolfer, H. & Harris, P. (1971). Host-specificity determination ofinsects for biological control ofweeds. Annual Review ofEntomology 16, 159-178 . -162-

APPENDIX A

6000000 •

• 5000000 •

y=109042 + 1972078 X 2 r = 0.984

2.0 2.5

concentration (ug/ul)

Fig. 1. Standard curve prepared from a hyoscyamine dilution series used to estimate the PA concentrations in aerial plant parts.